623267
research-article2015
PSIXXX10.1177/1529100615623267Rayner et al.How Do We Read, and Can Speed Reading Help?
So Much to Read, So Little Time: How
Do We Read, and Can Speed Reading Help?
Psychological Science in the
Public Interest
2016, Vol. 17(1) 4­–34
© The Author(s) 2015
Reprints and permissions:
sagepub.com/journalsPermissions.nav
DOI: 10.1177/1529100615623267
pspi.sagepub.com
Keith Rayner1†, Elizabeth R. Schotter1, Michael E. J. Masson2,
Mary C. Potter3, and Rebecca Treiman4
1
Department of Psychology, University of California, San Diego; 2Department of Psychology, University
of Victoria; 3Department of Brain and Cognitive Sciences, Massachusetts Institute of Technology; and
4
Department of Psychological and Brain Sciences, Washington University in St. Louis
Summary
The prospect of speed reading—reading at an increased speed without any loss of comprehension—has undeniable
appeal. Speed reading has been an intriguing concept for decades, at least since Evelyn Wood introduced her Reading
Dynamics training program in 1959. It has recently increased in popularity, with speed-reading apps and technologies
being introduced for smartphones and digital devices. The current article reviews what the scientific community knows
about the reading process—a great deal—and discusses the implications of the research findings for potential students
of speed-reading training programs or purchasers of speed-reading apps. The research shows that there is a trade-off
between speed and accuracy. It is unlikely that readers will be able to double or triple their reading speeds (e.g., from
around 250 to 500–750 words per minute) while still being able to understand the text as well as if they read at normal
speed. If a thorough understanding of the text is not the reader’s goal, then speed reading or skimming the text will
allow the reader to get through it faster with moderate comprehension. The way to maintain high comprehension
and get through text faster is to practice reading and to become a more skilled language user (e.g., through increased
vocabulary). This is because language skill is at the heart of reading speed.
Keywords
reading, speed reading, eye movements, rapid serial visual presentation, word recognition, comprehension
Introduction
One day in 2007, six-time World Speed Reading Champion
Anne Jones sat down in a popular bookstore on Charring
Cross Road, London, and devoured the latest Harry Potter
book in about 47 minutes (World Speed Reading Council,
2008). That worked out to a reading rate of over 4,200
words per minute (wpm). She then summarized the book
for some British news sources. Another speed-reading
enthusiast and promoter, Howard Berg, professes to be able
to read as many as 30,000 wpm (World's Fastest Reader on
Pelosi Bill, 2011). Reading rates of this kind seem extraordinary, given that college-educated adults who are considered
good readers usually move along at about 200 to 400 wpm.
Given the immense volume of text available to us on
a daily basis, it is unsurprising that most people would
like to increase their reading rates to that of Jones or
Berg. But is this possible? Some people suggest that it
is: Proponents of speed-reading courses claim that we
can dramatically increase our reading speed without
sacrificing our understanding of the content by learning
to take in more visual information at a single glance
and by suppressing the inner speech that often occurs
when we read silently. And now that text can be presented more dynamically, on digital devices as opposed
to paper, there are claims that new methods of text
presentation can allow us to read more quickly and
with good understanding. The most popular of these
technologies presents words rapidly one at a time on a
computer screen using what is called rapid serial visual
presentation (RSVP). The claim is that, freed from the
need to move our eyes, we can read more quickly than
we normally would. Other technologies manipulate the
Corresponding Author:
Elizabeth R. Schotter, Department of Psychology, University of
California, San Diego, La Jolla, CA 92093-0109
E-mail: eschotter@ucsd.edu
†In memoriam.
How Do We Read, and Can Speed Reading Help?
colors of presented lines of text, claiming that this can
help us to reduce skipping or repetition of lines.
Is there a unique form of reading in which speed and
comprehension are both high? Can we learn to read in
this way through speed-reading courses, or can we
achieve it with little or no practice by using special technologies? In this article, we address these questions. We
begin by reviewing psychological research on normal
reading and then discuss and evaluate methods that aim
to increase reading speed. We adopt this approach
because we believe that it is important to understand the
visual and mental processes that occur in typical silent
reading before determining whether special training or
technologies can allow us to increase speed without sacrificing comprehension. Therefore, the first section of this
article will review research on normal silent reading,
focusing on those research findings that are most important for evaluating claims about speed reading. We will
then consider the research on RSVP, the procedure that is
used by some currently popular speed-reading technologies. With this background in place, we will evaluate
speed-reading courses and technologies. As we will see,
research shows that there is not a unique and easily
learned behavior in which reading speed and comprehension are both high. There are effective ways of skimming a text to quickly find specific information, and
skimming may suffice if complete and detailed comprehension of the text is not a priority. Reading can also be
improved through practice. However, there is no quick
and easy procedure that can allow us to read a text more
quickly and still understand it to an equivalent level as
careful reading. You will see why this is the case as we
review what is known about normal reading.
Reading, Skimming, and Speed
Reading
Before we embark on a discussion of research on normal
reading and its implications for speed reading, we must
provide a definition of reading. Reading is typically defined
as the processing of textual information so as to recover
the intended meaning of each word, phrase, and sentence.
Of course, there are some forms of literature in which the
author intentionally provides some level of ambiguity. For
the most part, though, authors would like their readers to
understand what they intend to communicate and to
understand all of the words in the text. Often, the goal of
reading is to learn something new, whether it is a fact from
a textbook, a story from a novel, or instructions from a
manual. Successful reading thus requires more than recognizing a sequence of individual words. It also requires
understanding the relationships among them and making
inferences about unstated entities that might be involved
in the scenario being described.
5
Reading can be contrasted with skimming, in which
the goal is to quickly move one’s eyes through the text to
find a specific word or piece of information or to get a
general idea of the text content. As we will discuss later,
skimming rates can be as much as two to four times
faster than those of typical silent reading. Comprehension
rates, however, are lower when skimming than when
reading, suggesting a trade-off between speed and comprehension accuracy.
Where does speed reading fall on the reading–skimming
spectrum? Our brief discussion of trade-offs between speed
and comprehension suggests that a reader cannot “have his
cake and eat it too,” in the sense that comprehension must
necessarily suffer if the reading process becomes more like
skimming. Indeed, we will see there is little evidence for a
unique behavior, such as speed reading, in which speed
and comprehension are both high.
The Reading Process
To understand whether reading can be dramatically sped
up while maintaining comprehension, it is important to
understand how reading normally occurs. In this section,
we review the visual and mental processes that are
involved in silent reading when it proceeds as it typically
does in educated adults, at a rate of about 200 to 400
wpm. Throughout the course of this discussion, it is
important to keep in mind that reading is based on language; it is not a purely visual process. Speech is the
primary form of language, and all human societies have
a spoken language. (Groups of people who are deaf
have developed languages that use the visual modality,
but the primary languages employed by people with normal hearing all use the auditory modality.) We begin
learning our spoken language as babies, and this process
does not require explicit instruction. Reading and writing,
which are relatively recent cultural inventions that are
used in only some societies, normally require explicit
instruction. That instruction begins around the age of 6 in
many societies, although there are variations across cultures. It takes many years for a child to become a proficient reader. At first, children often read aloud, converting
a printed text into the more familiar spoken form.
Children gradually improve in their ability to read silently.
Even in cultures in which reading is highly valued and
widely taught, some children and adults never become
good readers. These observations indicate that speech is
the basic form of language; reading and writing is an
“optional accessory that must be painstakingly bolted on”
(Pinker, 1997, p. ix). Even though reading can be considered “an unnatural act” (Gough & Hillinger, 1980), many
adults in modern societies perform this act quite skillfully. The body of research we now discuss has shed light
on how they manage to do so.
6
Rayner et al.
Fig. 2. The visual forms of the same word (printed) displayed in two
different typefaces. Note that these two words appear to be physically
different (e.g., thick, straight lines vs. thin, slanted lines) but are nonetheless recognizable as the same word form.
Fig 1. Diagram representing the orthographic (letter-based), phonological
(sound-based), word, and conceptual representations of bag.
Writing systems
A logical starting point for reviewing the mental and
visual processes involved in reading is to consider what
the eye takes in and what the cognitive system must then
process: the elements of writing system. Writing normally
takes the form of visible marks on a surface, whether it is
a clay tablet, a sheet of paper, a computer monitor, or
another digital screen. In most languages, written words
are composed of smaller visual units that can be combined in various ways. The basic written symbols of
English and other alphabetic writing systems are letters,
which approximately represent the sounds of the language, the phonemes (Fig. 1). For example, the word bag
is represented by using a letter that represents the “b”
sound, a letter that represents the “a” sound, and a letter
that represents the “g” sound. The letters are arranged left
to right along a horizontal line in the case of English, but
other writing systems use other arrangements. For example, Hebrew arranges its symbols horizontally from right
to left.
In some writing systems that are not alphabetic, the
individual symbols that are arranged along a line represent units of meaning, morphemes, rather than units of
sound. For example, the Chinese character 人 stands for
“person.” Some Chinese characters contain components
that provide a hint about the character’s pronunciation.
However, these hints are not always present and, when
present, are not always consistent or helpful. Most modern Chinese words contain more than one unit of meaning and must be expressed by a sequence of characters.
For example, when the characters meaning “ground” (地)
and “board” (板) are written together, they convey the
meaning “floor” (地板). In Chinese, there are no spaces
between any characters, so the only way a reader knows
which two characters go together in a word is through
experience. For the most part, however, this does not
cause a problem for skilled readers.
The symbols of modern writing systems require no
shading, no color other than that needed to distinguish
the writing from the background, and no meaningful distinctions between lighter and darker lines or between
wider and narrower lines. Once a symbol has been
learned, it can be recognized in many printed forms. This
use of abstract codes rather than visual templates
(McConkie & Zola, 1979) helps to explain why we are
able to recognize that a word has the same meaning
regardless of the font or case in which it is printed (see
Fig. 2). Despite the fact that letters are recognized by
their abstract forms rather than visual templates, good
visual acuity is required to pick up the critical differences
between the marks that distinguish visually similar letters—for example, the difference between h and n. If this
difference is not perceived, one could mistake hot for
not.
Because discerning the correct visual form of words is
an important precursor to the rest of the reading process,
we now turn to a discussion of visual processing and the
role of eye movements in reading before moving on to
deeper aspects of the reading process: identifying words
and understanding meaning.
Visual processing and eye movements
Given that writing is composed of fine lines and marks,
the acuity (visual-resolution) limits of vision are an
important constraint on the reading process. The premise
behind some speed-reading courses is that it is possible
to use peripheral vision to simultaneously read large segments of a page, perhaps even a whole page, instead of
one word at a time (Brozo & Johns, 1986). However, such
a process is not biologically or psychologically possible.
One indication that it is not possible is that visual acuity
How Do We Read, and Can Speed Reading Help?
7
Fig. 3. Diagram illustrating visual acuity across the three regions of the visual field. Acuity
decreases continuously as a function of distance from fixation location. The visual field consists
of the fovea (center, with highest acuity), parafovea (middle region, with moderate acuity), and
periphery (farthest region from fixation, with lowest acuity).
is limited and that these limitations are what cause readers to make eye movements. Acuity is much higher in the
fovea (from the center of vision—the fixation location—
to 1° of visual angle away from it in any direction) than
in the parafovea (1°–5° away from the center of vision)
or periphery (areas more than 5° away from the center of
vision; Balota & Rayner, 1991; Fig. 3). To get a sense of
how small the foveal viewing area is, note that it is
roughly equivalent to the width of your thumb held at
arm’s length from your eye.
Saccades (quick, ballistic eye movements) allow readers to move the fovea to the word they wish to process
with the highest efficiency. Therefore, the oculomotor
(eye movement) system controls the sequence and timing of the visual system’s access to the text. Decisions
about how long to fixate a word and when to move the
eyes to the next word are to a large extent under the
control of cognitive processes (Rayner, Liversedge, White,
& Vergilino-Perez, 2003; Reingold, Reichle, Glaholt, &
Sheridan, 2012; see Rayner & Reingold, 2015). They are
not preprogrammed before one begins to read a text in
the way that a metronome is set by a musician before he
or she begins to practice a piece of music. This momentby-moment control helps to ensure that the next word
enters the system through foveal (high-resolution) vision
with the optimal timing. In the following sections, we
detail how the visual and oculomotor systems support
and constrain the reading process.
The visual system. As mentioned, acuity is highest in
the fovea, and this is the area in which the majority of
word recognition occurs. One reason that acuity is higher
in the fovea than in other areas of the visual field is related
to the distribution of two types of neural receptors that
respond to light—rods and cones. Cones are sensitive to
color and detail and are more effective in bright light,
while rods are sensitive only to brightness (e.g., shades of
gray) and motion and are mostly sensitive (i.e., useful) in
dimly lit rooms or at night. Cones are concentrated in the
fovea and decrease in density with increasing distance
from fixation. Rods are least concentrated in the fovea and
increase in density with increasing distance from fixation
(Fig. 4). Because cones are more sensitive to detail, this
means that acuity is higher in the fovea, where there are
more cones, than in nonfoveal areas.
Another reason why acuity is higher in the fovea than
in other areas of the visual field has to do with the way
information is transmitted from rods and cones (located
in the retina—a membrane lining the back of the inside
of the eyeball) to the brain. Information from rods is
pooled (averaged across a group of rods) before being
relayed to the brain, while information from individual
cones is relayed directly, without being combined with
information from other cones (Fig. 5). The consequence
of this organization is that even minute variations in the
Fig. 4. Diagrams showing the distribution of rods and cones across
the visual field (left panel) and retina (right panel).
8
pattern of light hitting the fovea will be preserved by the
cones. If one cone receives bright light and an adjacent
cone receives very dim light, the brain will perceive a
light/dark boundary. In contrast, minute variations in the
pattern of light hitting nonfoveal areas (where cones are
sparse and there are mostly rods) will be obscured. If one
rod receives bright light and an adjacent rod receives
very dim light, the brain will perceive a gray blob.
These facts about rods and cones have some important implications for reading. As we have discussed, all
text, regardless of writing system, is composed of combinations of lines, normally dark on a white background.
Therefore, fine discrimination between dark and light
areas is essential to recognizing the visual elements of
writing. If the light pattern coming from a word hits the
fovea, the cones will easily recognize such fine detail and
relay the pattern—with high fidelity—to the brain.
However, if the word hits nonfoveal areas and is sensed
primarily by rods, it will be relayed to the brain as an
average and will appear fuzzy. This will make it difficult
to discern the exact identities of the symbols (see the
beginning and the end of the sentence represented in
Fig. 3). In fact, when people are asked to report the identity of a word that is presented so briefly that they cannot
make an eye movement, accuracy is high in the fovea but
drops off dramatically outside of it, with performance
reaching chance level around the middle of the parafovea (approximately 3° of visual angle away from fixation;
Bouma, 1973; see also Bouma, 1978; Rayner & Morrison,
1981). These facts cast doubt on suggestions from speedreading proponents that people can read more effectively
by using peripheral vision, taking in an entire line or
even an entire page at a time.
Fig. 5. Diagram showing how information is relayed from the retina to
the brain. Information from cones (top panel) is relayed directly through
bipolar cells to ganglion cells and onward to the brain, whereas information from rods (bottom panel) is pooled (averaged) through bipolar
cells and ganglion cells.
Rayner et al.
Eye movements. For over a century, researchers have
been monitoring eye movements in order to study the
cognitive processes underlying reading. The technology
has evolved over the years so that eye tracking can now be
achieved with a high-speed video camera connected to a
computer and can be used to study readers of any age. In
general, these technologies work by computing the location of the eye up to one thousand times per second,
allowing the researcher to know, with precision to the millisecond, which word and where in the word the reader is
looking at a particular time. This information can then be
separated into times when the eyes remain in the same
location (i.e., during fixations) and times when they move
between locations (i.e., during saccades). In this section,
we review what has been discovered in studies of experienced adult readers, generally college students, who are
reading silently in their first language.
As mentioned earlier, the reason readers make saccades is to move their fovea to the next word. The eyes
are relatively stable during fixations, which last approximately 250 ms for experienced adult readers. In general,
no new visual information is obtained during saccades
(Matin, 1974), but cognitive processing continues during
this time (Irwin, 1998). This is important, because some
speed-reading technology developers have claimed that
saccades waste time. However, because cognitive processing continues during saccades, this time is not wasted.
We can conclude, however, that fixations are the reader’s
opportunity to obtain new visual information from the
text. Although the average fixation lasts about 250 ms,
there is considerable variability in how long an individual
fixation lasts. These variations reflect such things as the
legibility of the text (e.g., light/dark contrast, filled-in or
removed spaces between words), linguistic difficulty (e.g.,
word frequency, predictability, ambiguity), properties of
the reader (e.g., age or reading skill), and task goals (e.g.,
reading, proofreading, skimming).
One reason fixations last as long as they do is that eye
movements are motor responses that require time to plan
and execute. For example, even in the simple task of
moving the eyes to a new stimulus that appears either to
the left or to the right of the eye’s current location, the
reaction time is in the range of 100 to 1,000 ms, depending on the stimuli and experimental conditions (see
Gilchrist, 2011). The reaction time in reading is in the
shorter end of this range, 150 to 200 ms, because the eyes
generally move in one direction (Becker & Jürgens, 1979;
Rayner, Slowiaczek, Clifton, & Bertera, 1983). Additionally,
a number of processes are being conducted at once:
Processing of the fixated word, planning to move the
eyes forward, and processing of the upcoming word
using parafoveal information all overlap in time. This
means that saccade latencies in reading can be shorter
than in simple saccadic reaction-time tasks because some
How Do We Read, and Can Speed Reading Help?
of the cognitive processing that leads to the decision of
when to move the eyes can occur before the fixation on
a particular word even begins.
On average, forward saccades when reading English
last about 20 to 35 ms and span the distance of 7 letters.
Saccade durations, like fixation durations, are variable.
But the variation is mostly determined by the distance
traveled rather than by the cognitive and linguistic variables that affect fixation durations (Rayner, 1998, 2009).
Saccades usually move from one word to the next word.
However, about 30% of the time, readers move past the
next word to the following one. These skips are more
likely to happen when the word is very short, extremely
frequent, and/or highly predictable from the prior context. The word the has these characteristics, and it is
skipped about 50% of the time or more (see Angele &
Rayner, 2013). Importantly, just because a word is skipped
does not mean that it was not processed at all. All major
theories of reading posit that word skipping is based on
at least partial recognition of the word from information
obtained in parafoveal vision and/or expectations about
the word’s identity. In fact, if readers are given passages
to read in which words that most people skip over are
omitted, comprehension suffers rather dramatically
(Fisher & Shebilske, 1985). This shows that readers are
actually processing many or most of the words they skip
over, along with the words they fixate. It also suggests
that every reader is unique in terms of the timing and
sequence of words he or she needs to directly look at in
order to read efficiently. What works for some people,
such as the initial readers in Fisher and Shebilske’s study,
who were able to choose which words they fixated, does
not work for others, such as the readers who got the
modified text. The implication is that speed-reading
devices that control the timing of word presentation may
not be ready to use “out of the box” but instead may
need to be tailored to each individual user based on how
that person would naturally process the text.
Not all saccades move forward to the next word in the
text (Fig. 6). A small proportion of eye movements result
in refixations on the same word. Refixations are most
9
common for long words, about 7 or more letters long in
English, for which the end part of the word may not fall
within the word-identification span (described below).
About 10% to 15% of the time, skilled readers make
regressions, moving backward in the text to a previous
word. Regressions are different from return sweeps—eye
movements that go from the end of one line of text to the
beginning of the next. Although return sweeps and regressions are both right-to-left movements in writing systems
that go from left to right, such as English, return sweeps
continue to move forward with respect to the progression
of the text, whereas regressions move backward.
Because return sweeps tend to be long saccades, there
is some error in where they land, sometimes requiring an
additional fixation to correct ( Just & Carpenter, 1980). In
general, though, these corrective saccades take half the
time that normal saccades take and do not disrupt the
reading process too much (in fact, readers almost never
notice them). Some color-based technologies for presenting text have recently been developed that aim to make
it easier to make return sweeps. However, as we will
discuss in more detail later, return sweeps and other
aspects of oculomotor control are generally not the difficult part of reading. Faulty language processing generally causes problems in eye movement programming, not
the other way around.
Regressions are more important than return sweeps
with respect to understanding reading because they constitute a deviation of the reader’s eye movements from the
normal progression of the text. Although some regressions
are made to correct for oculomotor error (e.g., the eye’s
landing too far past the intended word), many regressions
are made to correct a failure in comprehension (e.g., when
the reader has misinterpreted the sentence). This is important in the context of speed-reading technologies that use
RSVP because these technologies do not allow people to
reread the text to correct misunderstandings in an intelligent way that is informed by the reader’s understanding of
the text. Given that most backward eye movements are
made in order to repair a failure in comprehension,
readers would maintain misinterpretations if they forced
Fig. 6. Schematic diagram of eye movements during reading.
10
Rayner et al.
themselves to keep moving forward and would comprehend the text less well (Schotter, Tran, & Rayner, 2014).
The use of foveal and parafoveal information to
process text. Now that we have reviewed the characteristics of eye movements, we return to the issue of how
information is obtained from a text once an eye movement is made. Visual perception occurs very rapidly—so
rapidly, in fact, that even if a word disappears completely
after being directly looked at for only 60 ms, reading
behavior is unaffected. That is, fixation durations have
been shown to be similar regardless of whether the word
remains visible or is erased or masked after 60 ms (Ishida
& Ikeda, 1989), and in both cases, the durations continue
to be sensitive to linguistic properties of the fixated word
(e.g., how common it is; Rayner et al., 2003). These findings suggest that visual perception takes up only a small
fraction of fixation durations, leaving time for higherlevel cognitive and linguistic processing to occur before
the decision of when to move the eyes next. Crucially,
the finding that readers’ eyes remained on words that had
disappeared for as long as they would have if the words
had still been there suggests that the reading system naturally delays looking directly at the next word until it has
performed a certain amount of linguistic processing of
the currently fixated word. Therefore, devices that present words faster than readers’ natural pace may run the
risk of presenting a word before the brain is prepared to
process and understand it.
Although acuity is lower in the parafovea than in the
fovea, information in the parafovea is not completely
ignored. If the word to the right of the fixated word disappears after 60 ms, reading behavior is disrupted (even
if the word reappears once it is directly fixated; Rayner,
Liversedge, & White, 2006). This finding suggests that
readers use information from more than just the fixated
word in order to read efficiently. This is important in light
of speed-reading technologies that present just one word
at a time: They do not allow the opportunity to use information from the next word.
A different way to determine how much is seen in a
single fixation is to examine eye fixations when readers can
see clearly only a limited window of text that moves as the
eyes move, using the gaze-contingent moving-window paradigm (McConkie & Rayner, 1975; see Rayner, 2014, for a
review). In this paradigm, the reader’s fixation location is
monitored and the text is manipulated based on the position of the eye (Fig. 7). The letters immediately surrounding
the center of fixation are revealed, forming a window of
clear text. Each of the letters outside the window, in some
studies including the spaces, is replaced with an x. On different trials, the window may be small (e.g., as small as a
single letter) or large (e.g., 40 letters). The experimenter
measures reading rate as a function of window size. The
Fig. 7. Diagram of the moving-window paradigm with a 9-character
window. From top to bottom, the lines show successive displays of
an example sentence. The centers of the blue rectangles represent the
locations of fixations, and the size of the rectangles reflects the size of
the window.
general finding is that reading rate increases as the window
size increases until it reaches an asymptote, the point at
which reading rate is equivalent to that with a completely
visible line. The window size at the asymptote point represents the size of the reader’s perceptual span. In general, in
English the size of the perceptual span is 3 to 4 letter spaces
to the left of fixation (McConkie & Rayner, 1976; Rayner,
Well, & Pollatsek, 1980) and 14 to 15 letter spaces to the
right of fixation (McConkie & Rayner, 1975; Rayner &
Bertera, 1979).
The perceptual span is determined not by physical
space or the distance traveled by the eyes but rather by
the amount of linguistic information that can be obtained
from the text. This fact is demonstrated in several ways.
First, the distance the eyes move during a saccade is
equivalent (when measured in terms of number of letters), regardless of the distance from the reader or size of
the text (Morrison & Rayner, 1981). Second, across languages, there are differences in the size and shape of the
perceptual span. In Chinese, it is 1 character to the left
and 3 to the right (Inhoff & Liu, 1998), and it is larger to
the left than to the right in both Hebrew (Pollatsek,
Bolozky, Well, & Rayner, 1981) and Arabic ( Jordan et al.,
2014). These differences reflect differences in the languages and writing systems. The perceptual span in
Chinese is smaller than in English because words are
shorter in Chinese (mostly 2 characters) than in English
(on average, about 5 letters). The size of the perceptual
span in the two languages is equivalent when measured
in number of words instead of number of characters, suggesting that there may be a generally optimal rate of
uptake of linguistic information that is more or
How Do We Read, and Can Speed Reading Help?
less consistent across languages. The asymmetry of the
perceptual span is reversed in Hebrew and Arabic compared to English because Hebrew and Arabic are read
right to left and English is read left to right—the perceptual span is larger in the direction of reading compared
to where the reader has already been. Third, the asymmetry of the perceptual span switches toward the direction of the eye movement when readers make regressions
(Apel, Henderson, & Ferreira, 2012). Thus, the size of the
perceptual span is not a physical limitation; otherwise, it
would not change with reading direction. Rather, the perceptual span reflects constraints on how linguistic information from the text is obtained and used to recognize
and understand words.
Further evidence that the perceptual span is limited by
cognitive and linguistic factors rather than just by acuity
comes from a study that used a clever manipulation to
compensate for the drop-off in acuity that occurs in the
parafovea (Miellet, O’Donnell, & Sereno, 2009). This
parafoveal magnification technique is similar to the moving-window technique. However, instead of text being
masked outside the center of vision, it is magnified as a
function of distance from fixation (Fig. 8). On every fixation, parafoveal and peripheral letters should have been
perceivable to the same degree as foveal letters because
of the magnification. It turned out that this manipulation
did not increase the size of the perceptual span. If our
ability to obtain useful information from the text were
limited solely by visual acuity, increasing the visibility of
eccentric letters would have allowed them to be perceived better and would have increased the size of the
perceptual span. The results of the study suggest that the
perceptual span is limited by our ability to identify and
process the meaning of the words in the text.
When a word falls within the perceptual span, it can
be perceived but not necessarily fully identified. In fact,
the area from which words can be identified, the wordidentification span, is much smaller than the perceptual
span. The word-identification span is about 7 characters
to the right of fixation in English (see Rayner, 1998).
Thus, with normal text, there is a small window comprising the currently fixated word and one or two words to
the right within which words are identified. The perceptual span is a larger window, used to perceive the visual
layout of the text (i.e., where the words and spaces are,
in writing systems that use spaces) in order to plan eye
movements. The final aspect of the perceptual span that
we should point out is that readers do not access any
information from the lines above or below the line currently being read (Pollatsek, Raney, LaGasse, & Rayner,
1993). This is important because it is inconsistent with
the speed-reading claim that someone can read an entire
page at once. The findings about the perceptual span
suggest that, contrary to the claims of speed-reading
11
Fig. 8. Depiction of the parafoveal-magnification paradigm. The location of each fixation is indicated with an arrow, and the corresponding
display for that fixation is represented. Consecutive lines represent the
chronological order of fixations. Reprinted from “Parafoveal Magnification: Visual Acuity Does Not Modulate the Perceptual Span in Reading,”
by S. Miellet, P. J. O’Donnell, and S. C. Sereno, 2009, Psychological Science, 20, p. 722. Copyright 2009 by the Association for Psychological
Science. Reprinted with permission.
courses, readers cannot obtain information from a very
large area of the visual field but rather primarily process
text in the center of vision (i.e., the fovea).
Further support for this idea comes from a study that
used the moving-mask paradigm (Rayner & Bertera,
1979). This technique is similar to the moving-window
technique except that the mask (i.e., the string of xs that
replaces the letters in the text) moves in synchrony with
the eyes, obscuring the text in foveal vision instead of
nonfoveal vision. In this study, readers were extremely
disrupted when even a single letter in the fovea was
masked—their reading rates dropped by half and continued to drop precipitously as the size of the mask increased.
Therefore, readers were not able to read effectively by
relying on parafoveal and peripheral vision alone.
Although research using the moving-window paradigm provides us with an assessment of how much of the
text readers can perceive on a given fixation, it tells us
little about what type of information is obtained from
words before they are fixated and how this information is
used in reading. To investigate this, researchers have
used the boundary paradigm (Rayner, 1975). Here, people read sentences in which a specific word is manipulated. Initially, it is replaced with a different word (or
nonword) called the preview. As illustrated in Figure 9,
there is an invisible boundary located just prior to this
preview word, which, when crossed by the reader’s gaze,
causes the preview to change to the target word—the
word that makes sense in the sentence. Readers are rarely
aware that words are changing because the change
occurs during a saccade when vision is suppressed
(Matin, 1974). However, the change may be noticeable if
Rayner et al.
12
Fig. 9. Diagram of the boundary paradigm. The vertical dashed line represents the location of the
boundary, which is invisible in the experiment. When the reader’s gaze (represented by gray circles)
crosses this location, between the third-to-last and second-to-last lines, the preview word (“kfevl,” in
this example) changes to the target word (“front”).
it is visually drastic (e.g., if an ascending letter such as h
changes to a descending letter such as y) or if the eyes
were very close to the preview word before the boundary was crossed (Slattery, Angele, & Rayner, 2011).
The general finding from studies using the boundary
paradigm is that readers obtain parafoveal preview benefit. That is, they are faster to read the target when the
preview was identical to the target than when the preview
differed from the target (for a review, see Schotter, Angele,
& Rayner, 2012). The boundary paradigm allows researchers to manipulate the nature of the relationship between
the preview and target in order to gain more detailed
information about exactly what type of information causes
the preview benefit. We turn now to studies that have
addressed this issue, which is important in the context of
speed-reading technologies that present one word at a
time and do not allow for parafoveal preview.
Readers obtain something more abstract from the
input than pure visual features. This was cleverly demonstrated by McConkie and Zola (1979), who had readers
learn to read something like “AlTeRnAtInG cAsE” and
then either changed the case of each of the letters (making the text read “aLtErNaTiNg CaSe”) or left the letters
unchanged during a saccade. Changing the case of the
letters did not affect fixation durations at all, suggesting
that readers had discarded the exact visual form of the
letters and that their reading behavior was based on
abstract letter codes (i.e., the identity of the letters; see
also Friedman, 1980; Rayner, McConkie, & Zola, 1980;
Slattery et al., 2011). But beyond the identity of letters,
there is abundant evidence that readers are faster to read
a target word when the preview was phonologically
related to it. This holds true for readers of alphabetic
writing systems such as English (Pollatsek, Lesch, Morris,
& Rayner, 1992) and French (Sparrow & Miellet, 2002) as
well as non-alphabetic systems such as Chinese (Pollatsek,
Tan, & Rayner, 2000). Although initial evidence suggested
that preview benefit does not reflect processing of word
meaning (Rayner, Balota, & Pollatsek, 1986; Rayner,
Schotter, & Drieghe, 2014), more recent evidence indicates that some aspects of the upcoming word’s meaning
are processed under certain conditions (Hohenstein &
Kliegl, 2014; Rayner & Schotter, 2014; Schotter, 2013;
Schotter, Lee, Reiderman, & Rayner, 2015).
Finally, research using the boundary paradigm has
demonstrated that readers typically do not obtain information from the second word to the right of the fixated
word (e.g., Angele & Rayner, 2011; Rayner, Juhasz, &
Brown, 2007) unless the two words following the fixated
word are short (i.e., close to the fovea), common, and/or
predictable (Cutter, Drieghe, & Liversedge, 2014; Radach,
Inhoff, Glover, & Vorstius, 2013). If observed, preview
benefits from the second word to the right of the fixated
word are typically smaller in magnitude than preview
benefits from the word immediately following the fixated
word.
The findings from both the moving-window paradigm
and the boundary paradigm are quite consistent with the
research we discussed earlier regarding the limitations on
reading and the fact that people need to move their eyes
so as to place the fovea over the region that they want to
process. That is, they suggest that people need to move
their eyes to look at words directly in order to read efficiently. These findings do not align with some of the
central claims of speed-reading courses, such as that
readers can obtain information from a large area of text
How Do We Read, and Can Speed Reading Help?
13
in a single fixation. Nor do they align with the claims of
some speed-reading technologies, such as that reading
will be more efficient if words are presented one at a
time, without the opportunity for parafoveal preview.
Reading speed across people. Is there variability in reading speed? Unsurprisingly, reading speed varies greatly
among individuals (Rayner, 1998; see Table 1), most notably
as a function of reading skill: Fast readers make shorter fixations, longer saccades, and fewer regressions than slow
readers (Everatt & Underwood, 1994; G. Underwood, Hubbard, & Wilkinson, 1990; see Everatt, Bradshaw, & Hibbard,
1998). Importantly, in a study that examined a large number
of cognitive skills, the factor that most strongly determined
reading speed was word-identification ability (Kuperman &
Van Dyke, 2011). This finding suggests that reading speed is
intimately tied to language-processing abilities rather than
eye movement–control abilities.
Models of eye movement control. While we have
learned a great deal about eye movements and reading
over the past few decades, there are still some open
questions. However, the research that has been done so
far has provided such a detailed understanding that
researchers have been able to develop sophisticated
models of eye movements during reading. These models
are computer programs that do a very good job of predicting how long readers will look at words and where
they will move their eyes next. There are a number of
such models (see Reichle, Rayner, & Pollatsek, 2003), but
two have received the most attention: the E-Z Reader
model (Reichle, Pollatsek, Fisher, & Rayner, 1998) and the
SWIFT model (Engbert, Nuthmann, Richter, & Kliegl,
2005). Although a full discussion of these models is
beyond the scope of the present article, the most important point for present purposes is that, in the models,
word identification is a primary determiner of when and
where readers move their eyes. We do not set an unchanging pace and length for eye movements in advance of
reading a text. Rather, we vary our pace and our movements depending on our ongoing cognitive processes:
how well we are processing the incoming information. A
second important point is that, in adapting such models
to account for the development of reading skill in children, the parameters that need to be modified are those
associated with language-processing ability rather than
those that specify eye movement–control ability (Reichle et
al., 2013). This finding from the modeling work further supports the idea that language processing, rather than the ability to control the movements of one’s eyes, is the primary
driver of reading performance. Given this, we now turn to
an overview of one major aspect of language processing:
word recognition, and its role in reading.
Word recognition
Reading obviously consists of more than recognizing
individual words. But, given that a writing system represents the words of a language, words are the basic stepping-stones to reading and comprehension. You cannot
reasonably expect to understand a text if you do not
know what the words mean (imagine trying to read an
unfamiliar language). In this section, we detail what is
known about word recognition.
One of the biggest influences on the time that it takes
to recognize a word is its frequency of occurrence—
essentially, how often the word has been encountered
before (Rayner, 1998, 2009). In reading, as in listening,
Table 1. Mean Fixation Duration, Mean Saccade Length, Percentage of Regressions, and Reading Rate for 10 Skilled
Readers (Rayner, 1998)
Reader
K. B.
J. C.
A. K.
T. P.
T. T.
G. T.
G. B.
B. B.
L. C.
J. J.
Mean
Fixation
duration (ms)
Saccade length
(characters)
Regressions (%)
Reading rate (words
per minute)
195
227
190
196
255
206
205
247
193
241
216
9.0
7.6
8.6
9.5
7.7
7.9
8.5
6.7
8.3
7.2
8.1
6
12
11
15
19
4
6
1
20
14
11
378
251
348
382
244
332
347
257
314
230
308
Note: Adapted from “Eye Movements in Reading and Information Processing: 20 Years of Research,” by K. Rayner, 1998,
Psychological Bulletin, 124, p. 393. Copyright 1998 by the American Psychological Association. Adapted with permission.
14
words that are more common (e.g., house) require less
time to recognize than words that are less common (e.g.,
abode). The effects of word frequency can be thought of
as operating through practice or experience; the more
times you have encountered and recognized a word, the
easier it will be for you to do it again in the future.
The word-frequency effect can be demonstrated for
written words seen in isolation by measuring reaction
time in tasks such as the lexical decision task (Stanners,
Jastrzembski, & Westbrook, 1975), the naming task
(Berry, 1971; Forster & Chambers, 1973), and the categorization task (Van Orden, 1987). These tasks are depicted
in Figure 10. In a lexical decision task, words and nonwords are briefly presented on a computer screen and
the participant’s goal is to determine whether each letter
string is a word or not. In a naming task, the goal is to
read the word aloud as quickly and as accurately as possible. In the categorization task, the goal is to determine
whether the word is a member of a certain semantic
category—for example, the category of foods. For words
embedded in passages of text, the frequency effect can
be measured by examining eye fixation times during
silent reading (Rayner & Duffy, 1986) or event-related
brain potentials (ERPs; King & Kutas, 1998) during the
reading of words presented one at a time via RSVP.
During silent reading and RSVP, the participant’s goal is
to understand the sentence in order to recall the text or
to answer comprehension questions about it. Despite
these differences in the requirements or goals of the
tasks (e.g., determining word status, accessing pronunciation, or accessing meaning), the frequency effect is
apparent across all of them (although the magnitude of
the effect changes depending on the task; Schotter,
Bicknell, Howard, Levy, & Rayner, 2014). Thus, it is a
ubiquitous and robust influence on the word-recognition process.
Rayner et al.
From letters to words. Recognizing a word, in an
alphabetic writing system, involves processing the letters
within it. As we discussed earlier, the letters in the words
of alphabetic writing systems symbolize the sounds that
are used to pronounce the word, and the pronunciation
is in turn associated with a meaning. We now turn to the
question of how people recognize the individual letters
within the word and retrieve the pronunciation and
meaning.
For most words, including short and medium-sized
ones, all of the letters within the word are recognized
simultaneously. For very long words such as antidisestablishmentarianism, it may not be possible to recognize
all the letters simultaneously (recall that the size of the
word-identification span is only about 7 letters to the
right of eye fixation in English). These words require
multiple fixations in order to be recognized. Simultaneous
recognition of letters is more efficient within words that
are well known than words that are completely new to
the reader. In fact, it is actually easier to recognize a letter
inside a known word than it is to identify it in isolation!
This effect, termed the word-superiority effect, was discovered over 125 years ago by Cattell (1886) and later
confirmed with more rigorous methods by Reicher (1969)
and Wheeler (1970). Experiments demonstrating this
effect have used the general design that is shown in
Figure 11.
A person sees either a word (e.g., “word”), a nonword
with the same letters in a jumbled order (e.g., “orwd”), or
a single letter (e.g., “d”) presented very briefly on a computer screen—so briefly that there is not enough time to
make an eye movement. Next, a mask stimulus (e.g.,
“####”) replaces the initial stimulus, barring the person
from holding onto the information about the word, nonword, or letter in any type of raw visual form. The person
is cued to a particular position in the stimulus (e.g., the last
Fig. 10. Diagram of three tasks commonly used to study word reading. For the lexical decision task, in which the goal is to decide if the letter
string is a word or nonword, the top panel shows a nonword trial and the bottom panel shows a word trial. For the naming task, the goal is to read
aloud the word. For the categorization task, in which the goal is to decide if the word fits into a given category (e.g., “foods”), the top panel shows
an item that fits the category and the bottom panel shows an item that does not.
How Do We Read, and Can Speed Reading Help?
Fig. 11. Diagram of the paradigm used by Cattell (1886), Reicher
(1969), and Wheeler (1970) to demonstrate the word-superiority effect.
From top to bottom, the first three panels show successive displays in
the experiment. The pair of panels at the bottom show the participant’s
potential responses.
position) and asked to report which of two possible letters
had been presented at that location (e.g., “d” or “k”). Note
that in this example (and for all the items in Reicher’s and
Wheeler’s experiments), either of the response options in
the word condition would make a real word (i.e., either
word or work). This means that using knowledge about
what letters would make the letter string form a word in
English could not help a person perform above the level
of chance, which is 50%. But performance in the word
condition is much better than chance and much better
than in either the nonword condition or (quite surprisingly!) the isolated-letter condition.
Even more impressive than the word-superiority effect
is the fact that, once established, the word-recognition
system is so efficient that it can interfere with processes
that seem to be very basic, such as color recognition. The
clearest demonstration of this is the Stroop effect, named
after the scientist who discovered it (Stroop, 1935) and
replicated hundreds of times since (see MacLeod, 1991,
for a review). This effect is so robust that it can be demonstrated easily in classrooms with a single piece of
paper and a timer.
In the Stroop task, the goal is to name the color of the
ink in which each word is printed as fast and as accurately as possible—you can try it yourself by using Figure
12. Start with the first column and time yourself. Then try
the second column, and then try the third. You may
notice that each column feels more difficult than the last
and takes you more time. This is because when the word
names a color that is different from the color you are trying to say, as in the third column, you have difficulty
15
selecting the correct response. You do not experience
such competition in the first column (because the word
and ink color lead to the same response) or in the second
column (because the string of x’s does not bring a particular word to mind).
What is so surprising about this effect, besides how
robust and easily demonstrated it is, is that we would generally think that color naming should be so easy that the
words should be irrelevant—recall that the cones (photoreceptors in your eye) are specifically tuned to detect
color. In fact, some recently developed speed-reading
apps that use color claim “Ever wonder why stop lights
use color and not words? It’s because the human brain
processes color very quickly—much more quickly than it
can process words” (http://www.beelinereader.com/). The
Stroop effect suggests that it might not be as simple as that;
once a reader is proficient in a language (this effect does
not appear when the words are printed in a language the
person does not know), word recognition is so incredibly
quick and strong that it can interfere with retrieval of a
competing word—the name of the color.
The optimal viewing position effect. It is important
to note that, with the Stroop effect, the interference created by a mismatch between the word name and the ink
color is strongest when people are looking at the center
of the word than when they are fixating more peripheral
Fig. 12. Example stimuli in the Stroop paradigm. The task is to name
the color of the ink that the word is printed in as quickly and accurately
as possible. Column 1 shows stimuli in the congruent condition (each
word names the color in which it is printed), Column 2 shows stimuli
in the neutral condition (each stimulus is a colored string of x’s), and
Column 3 shows stimuli in the incongruent condition (the words name
colors different from the colors they are printed in).
16
letters (Perret & Ducrot, 2010). This is important because
of another robustly demonstrated effect—the optimal
viewing position (OVP) effect—in which words that are
presented in isolation are recognized more efficiently
when people are looking in the center of the word than
when they are looking at more external letters (O’Regan,
Lévy-Schoen, Pynte, & Brugaillère, 1984). Although this
effect is consistently obtained for single-word presentations, it is small, on the order of only a few milliseconds.
It is even smaller during natural reading of text (Vitu,
O’Regan, & Mittau, 1990), probably because of the ability
to obtain parafoveal preview that we discussed earlier.
This will become an important point later when we talk
about some recent speed-reading technologies that present words centered near the OVP. The claim is that this is
done because people cannot recognize a word until they
fixate the OVP. The research shows, however, that people
are not incapable of recognizing words if they do not fixate the OVP; instead, the recognition process is just
slightly slowed.
The role of phonology in silent reading. Given that
all writing systems represent words and given that the
primary form of language is vocal and auditory, not
visual, it is no surprise that phonological processing plays
an important role in reading—even in silent reading (see
Leinenger, 2014). This is true not only for alphabetic writing systems, in which letters represent the individual
phonemes in words, but also for non-alphabetic writing
systems such as Chinese. We will now describe a number
of studies showing that readers naturally access the
sounds of words while reading silently and that attempts
to inhibit this process generally have negative effects on
reading. These studies are important to examine in light
of the speed-reading claim that use of inner speech
causes problems for readers.
First, consider a study in which skilled readers were
asked to indicate (with a button press) whether briefly
presented words were members of a certain category
(e.g., foods; Van Orden, 1987). The people in this study
incorrectly responded “yes” to a word that was not a
member of the category (e.g., meet for the category food)
about 19% of the time if that word was a homophone of
a true member of the category: In this example, meet is
pronounced the same as meat, which is a food. Incorrect
“yes” responses to non-homophone words—for example,
melt—occurred only 3% of the time. This is because
while melt shares the same number of letters with meat
as meet does, it does not share the pronunciation. When
people did correctly respond that, for example, meet is
not a food, it took them longer to do so than it did for
melt. These results suggest that people accessed the
sounds of the word and were strongly influenced by the
fact that its phonological code was identical to that of an
Rayner et al.
actual member of the category. Readers of Chinese also
activate the sounds of words when they read (Xu,
Pollatsek, & Potter, 1999), although Chinese is not an
alphabetic writing system.
Studies using homophones have also found evidence
for inner speech during silent reading of text. For example, when people were asked to read silently and make
very quick judgments about whether sentences made
sense, they incorrectly accepted a sentence like “She has
blonde hare” about 12% of the time, more often than they
incorrectly accepted a sentence like “She has blonde
harm” (about 5% of the time; Treiman, Freyd, & Baron,
1983). If use of phonology were an unimportant process
that readers could easily turn off, we would have expected
them to do so. The fact that they could not suggests that
phonology is fundamental to the process of reading.
What would happen if we eliminated inner speech
during reading? It is difficult to design a procedure that
prevents people from performing a mental activity, but
researchers have tried two approaches. One approach is
to use biofeedback techniques to get people to reduce
the activity that sometimes occurs in their speech muscles while they are reading silently. In one study, a tone
was played when a monitor picked up such activity and
readers were instructed to keep the tone off (Hardyck &
Petrinovich, 1970). Readers were able to keep the tone
off to some extent, and their performance on a later comprehension test was unaffected if the text was easy but
was impaired if the text was more difficult. This effect did
not just seem to reflect the added complexity of reading
while listening for a tone, because comprehension did
not suffer in a group of participants whose arm muscle
activity was monitored in a similar way.
Another way to try to decrease the use of inner speech
during silent reading is to have people count aloud or
repeat an irrelevant word or phrase while they are reading (Daneman & Newson, 1992). In such studies, readers
showed poorer comprehension of passages when they
repeated the irrelevant words, but not when they repeated
an irrelevant sequence of taps with their fingers. This latter result suggests that it was speech, rather than the
requirement to perform two tasks, that impaired reading
comprehension. In another study in which people
repeated irrelevant speech while reading silently, readers
were quite good at answering comprehension questions
about individual concepts but were worse at answering
questions that required them to combine concepts and
make inferences (Slowiaczek & Clifton, 1980). These
findings support the idea that, when it comes to understanding complex materials, inner speech is not a nuisance activity that must be eliminated, as many
speed-reading proponents suggest. Rather, translating
visual information into phonological form, a basic form
of language, helps readers to understand it.
How Do We Read, and Can Speed Reading Help?
Word-recognition models. Earlier, we discussed how
researchers have developed detailed models of eye
movement control. There are also detailed models of the
word-recognition process. The models that have received
the most attention and motivated the most research are
the dual-route cascaded (DRC) model (e.g., Coltheart,
Rastle, Perry, Langdon, & Ziegler, 2001) and parallel distributed processing (PDP) models, also called connectionist or triangle models (e.g., Seidenberg & McClelland,
1989). The models disagree on some points, showing
that scientists still have things to learn about the process
of recognizing words. However, there are some important areas of agreement. For one, the models share the
basic assumption that bottom-up orthographic input
interacts with what people know about words to produce pronunciations and meanings. That is, both detailed
visual information from the text and prior experience
with the language are necessary. Another area of agreement between models of word recognition is that internal
representations of phonology play a role in the recognition of words that are presented visually and not pronounced aloud by the reader.
The role of context in selecting the correct word
meaning. In some cases, a word has two separate
meanings but the same spelling and the same pronunciation. An example of such a homograph is bank, which
can refer to a financial institution or the side of a river.
Lexical ambiguity of this sort complicates word recognition. Lexical ambiguity is not rare—the English language
contains over 1,500 homographs (Leinenger & Rayner,
2013)—and it may even be a desirable feature of language because it allows us to reuse word forms for multiple purposes instead of creating new, increasingly large
and complex words (Piantadosi, Tily, & Gibson, 2012).
Still, difficulty arises because merely knowing the word’s
visual form and pronunciation does not indicate which of
the two meanings the reader should interpret.
When a homograph is encountered in isolation, it is
impossible to discern its intended meaning. Luckily,
readers usually have the benefit of a sentence context
that can give them clues as to which meaning they
should interpret. For example, it would be easy to infer
that bank means “financial institution” when it is preceded by “John went to deposit some money at the . . .”
In contrast, you might be more likely to infer that it
means “side of a river” if it were preceded by “John went
to sail a boat at the river . . .” In this example, the information that helps distinguish the meaning of bank
comes before the word itself. But sometimes the disambiguating information appears after the word. In these
situations, readers are less likely to make the correct initial interpretation. Rather, readers tend to make regressions back to the word if they had initially misinterpreted
17
it (e.g., Duffy, Morris, & Rayner, 1988; Rayner & Duffy,
1986).
Difficulty or uncertainty discerning exact word meaning can also arise when word forms are visually very
similar, as with orthographic-substitution neighbors (e.g.,
birch and birth) and transposition-letter neighbors (e.g.,
calm and clam). The likelihood that the word will be
misinterpreted is higher if the correct word is rare and
the highly similar word is more common. Context can
help readers in such cases; readers have an easier time
recognizing the correct word (as opposed to the visually
similar word) when the preceding sentence context indicates the correct meaning ( Johnson, 2009; Slattery, 2009).
Sentence context also helps readers with words that
are neither homographs nor neighbors of another word,
indicating that the effect is fairly general. For example, a
constraining context, one that makes a particular word
predictable, leads to that word’s being fixated less often
(i.e., skipped more) and for less time than if it were preceded by a neutral context that could be completed by a
wide range of words (Ehrlich & Rayner, 1981; Zola, 1984;
see Rayner, 1998, 2009, for reviews). A constraining context also speeds response time in naming (Stanovich &
West, 1979, 1981; West & Stanovich, 1982) and lexical
decision tasks (Schuberth & Eimas, 1977) and affects the
brain’s response to words as measured with ERPs (Kutas
& Hillyard, 1984; see Kutas & Federmeier, 2011). The
findings about the importance of context in visual word
recognition suggest that training courses that teach students not to reread and speed-reading devices that do
not allow for going back will make readers more likely to
misinterpret words when they occur in less constraining
contexts.
Comprehension
Given that reading typically progresses rather smoothly,
we may not notice the subtle complexities that are associated with understanding the meaning of a text. We have
already mentioned that some words are ambiguous
because they have multiple meanings. Sentences also can
be ambiguous in that it is not always obvious how each
individual word in a given sentence functions and relates
to the other words (i.e., the sentence’s syntax). Consider
the sentence “While Mary was mending the sock fell off
her lap.” When reading a sentence like this, readers often
initially misinterpret it (the term for this is getting garden
pathed; see Frazier & Rayner, 1982; Rayner, Carlson, &
Frazier, 1983). This happens because readers, like listeners, tend to process words incrementally, as soon as they
have identified them. They often assume at first that “the
sock” is the object of “was mending” (i.e., that what Mary
was mending was the sock). Of course, after reading the
end of the sentence, it is clear that that interpretation
18
could not possibly be correct (if so, then what fell off her
lap?). In reality, the sentence should be interpreted such
that “the sock fell off her lap” is a sentence unit (a constituent) and that Mary was mending something that was
not stated in the text. When we introduce this idea and
these types of sentences to students in the classroom,
they often note that ambiguity could be avoided if the
writer simply put a comma after “was mending” to disambiguate the sentence by separating the two constituents.
One might argue that good copy editors would not let
such a sentence occur and that such poorly written sentences should be rare. However, the comma is indeed
optional in English grammar and writers (and speakers)
are not very good at anticipating whether readers (or
listeners) are likely to misinterpret what they write (or
say; Ferreira & Schotter, 2013). Therefore, garden-path
sentences like this occur much more often than one
might imagine (they are quite notorious for occurring in
newspaper headlines; e.g., “McDonald’s fries the holy
grail for potato farmers”: Crash Blossoms Up the Garden
Path, 2009).
Further complicating the process of understanding
sentences is that words can be ambiguous with respect to
their grammatical role—that is, their part of speech. In
the garden-path example we have been discussing, the
problem in interpreting the sentence is not due to difficulty discerning the meaning of the individual words but
rather to difficulty deciding whether Mary was mending
the sock or whether the sock fell. Add to that ambiguity
the difficulty of deciding whether a word itself means
one idea or another and you find yourself in a very difficult situation. For example, consider a sentence that
begins, “The desert trains . . .” With only these words,
you will not be able to figure out how this phrase
should be interpreted until you get more information
from the rest of the sentence. Context would help you
determine whether those words should be interpreted
as a noun and verb (i.e., “The desert trains boys to be
men”) or an adjective and a noun (i.e., “The desert
trains are hot and dusty”). The information that would
help you understand the roles and meanings of “desert
trains” occurs after the ambiguous phrase. When you
first encounter the words “desert trains,” you must make
a best guess as to how to understand them. When your
first guess at an interpretation is wrong, then you run
into problems. In these situations, readers fix the problem by making regressions and rereading the sentence
(or parts thereof) in order to find the correct interpretation (Frazier & Rayner, 1987).
Fixing comprehension failure. The examples we
have been discussing show that discourse can sometimes
contain syntactic ambiguities and it is only via careful
reading that the reader will be able to appreciate the
Rayner et al.
appropriate meaning. Readers are often able to fix comprehension problems by rereading—something that
speed-reading training courses discourage and that
devices using RSVP may make impossible. In fact, recent
research demonstrates this point quite clearly. Schotter,
Tran, and Rayner (2014) had people read garden-path
sentences (such as the one in Fig. 13) while their eye
movements were monitored. The sentences were presented either normally or with a trailing-mask manipulation, in which each letter in a word was replaced with an
x after the reader’s eyes moved past it. This manipulation
ensured that readers had only one encounter with the
word—if they returned to reread it, they would be looking at a string of x’s. Schotter and colleagues found that
readers could accurately respond to a two-alternative
comprehension question about the sentence much better
when they were able to reread words (i.e., when they
made a regression in the normal reading condition: 75%
accuracy) than when they could not (i.e., when they
made a regression in the trailing-mask condition: around
50% accuracy—chance performance). These findings
suggest that having only one encounter with the words in
the sentence might not be sufficient for successful understanding of the text.
During silent reading, the eyes slow down slightly at
the ends of sentences or phrases ( Just & Carpenter, 1980;
Warren, White, & Reichle, 2009). However, this does not
mean that all comprehension occurs there. Readers have
already understood most of the words and sentence
meaning by that point. Evidence for this idea comes from
the fact that people make occasional regressions to
reread prior parts of a sentence and that these regressions are not triggered only from the punctuation mark at
the end of the sentence; rather, they can be triggered
from various parts of the sentence (von der Malsburg &
Vasishth, 2013). This finding suggests that readers noticed
a failure in comprehension partway through reading a
sentence. Some wrap-up comprehension processes do
occur at the ends of phrases or sentences ( Just &
Carpenter, 1980), but much language processing occurs
as the individual words come in.
Comprehension of passages of text. So far, we have
talked about comprehension primarily at the level of
individual words and sentences. But readers must also
build an ongoing mental model of the entire text, integrating information across sentences (e.g., Zwaan & Radvansky, 1998). They must not only understand each
sentence, which requires short-term memory, but also
retain the information in longer-term memory. Sometimes
readers make very long regressions—for example, from
the end of a paragraph to the beginning—in order to
make sense of the current discourse’s reference to something that was stated earlier. The important point here is
How Do We Read, and Can Speed Reading Help?
19
Fig. 13. Diagram of the trailing-mask paradigm (Schotter, Tran, & Rayner, 2014). From top to bottom, the lines show successive displays of an
example sentence. Gray circles represent the locations of fixations. Once a reader made a forward saccade out of a word, each of its letters was
replaced with an x for the remainder of the trial, even if the reader looked back to it, as shown in the second-to-last line.
that one advantage the reading of normal text has over
methods that force the eyes to move straight down a
page or present words one at a time is this opportunity—
the opportunity to move backward in the text in order to
recover information that was initially missed or
forgotten.
Measuring comprehension. It is fairly easy to measure reading rate by computing the number of words
read in a set period of time (e.g., wpm). It is harder to
measure comprehension. Given that speed-reading
courses and devices promise to increase speed without
decreasing comprehension, the question of how comprehension should be measured is critical to discuss.
Most often, researchers measure comprehension with
multiple-choice questions that are presented after a sentence or a paragraph that probe the reader’s memory for
it. The advantage of this method is that it is fairly easy to
get a score for comprehension. However, the test is only
as good as the questions and the incorrect answers that
are provided as foils for the correct answer. In some
cases, it is possible for people who never read the text to
do surprisingly well on the test (although they would
probably have a hard time doing better than someone
who read the text carefully). Furthermore, there is evidence that reading behavior changes depending on the
type of questions included on a comprehension test
(Radach, Huestegge, & Reilly, 2008; Wotschack & Kliegl,
2013), suggesting that readers may adopt different strategies for reading in anticipation of what they need to
retain from it.
An alternative to multiple-choice questions are tests
requiring some open-ended response, such as asking
readers to give a summary of what was just read. This
type of test may provide a better estimate of comprehension, but this method is not without its pitfalls. Scoring
open-ended summaries is much more subjective than
scoring multiple-choice questions, and comprehension
scores for the same response can vary drastically depending on who scores it, how rigorous the coding scheme
was, and so on. Although some variability in response
scoring can be eliminated by using a consistent scorer,
such as a computer program, we are still a long way from
creating a computer program with a native-like grasp on
all the nuances of natural language.
Participants in speed-reading courses are often evaluated for comprehension both before and after the course,
not for research purposes but to show participants how
their reading has changed. Prior to beginning a speedreading program, students are typically given a pretest that
measures both their reading speed and comprehension.
At the end of the training program, they are given a posttest. Sometimes the pretest is harder than the posttest, and
other times trainees are tested repeatedly on the same
material (Carver, 1971, 1972). In both cases, it is inevitable
that their performance will be better on the posttest
merely because of the relative difficulty of the tests or
because of repeated exposure. In other cases, training
programs change the measurement used to assess reading
between pre- and posttest. A new student in a speedreading course generally has his or her reading speed
measured in a straightforward way (e.g., wpm). However,
Rayner et al.
20
after the course, what is often measured is the Reading
Efficiency Index (RE Index; Carver, 1990), based on the
argument that rapid reading rates should be qualified by
the percentage of the material that the reader is able to
comprehend. To calculate this measure, the reading rate
is multiplied by the percentage of correct responses on
the comprehension test. Thus, if a reader had a reading
rate of 5,000 wpm and scored 60% on the comprehension
test, his or her RE Index would be 3,000 wpm (5,000 ×
0.60). Since comprehension was not factored into the
measurement of the pretest assessment, trying to infer
what type of improvement the course caused relies on
comparing apples and oranges. For example, if a multiple-choice test with four questions is used, people who
“read” at 5,000 wpm and understood absolutely nothing
(i.e., answered the comprehension questions at the level
of chance—25%) would get an RE Index score of 1,250
wpm. But if they did not understand anything about the
text, can we really say that they were reading?
Conclusions about the natural
reading process
We have seen that reading is an elegantly choreographed
dance among a number of visual and mental processes.
Modern research has shown that, contrary to some earlier
views (Goodman, 1967), reading is not a psycholinguistic
guessing game in which we guess the identities of words
and other linguistic units based on minimal visual input.
Rather, we pick up detailed visual information from the
text, moving our eyes so that we fixate most words once
and going backward to reread if problems arise. The
visual information that we obtain, combined with our
knowledge of the language we are reading, allows us to
identify the words in the text and to comprehend it.
Reading silently is faster for skilled readers than either
reading aloud or hearing someone else read the text.
This difference reflects, in part, limitations on how
quickly we can talk. A speaking rate of 150 to 160 wpm
is comfortable; this is the rate that is recommended for
speakers who are recording audiobooks or podcasts.
Those who have practiced rapid speaking, such as auctioneers, can maintain rates close to that of skilled reading, between 250 and 400 wpm. Some people prefer to
double the speed of audio books to get through them
more quickly, but this still brings the speed up only to the
middle of the average reading-rate range (i.e., around
300 wpm). Viewed in this light, most people reading this
article are already going quite quickly. Yet in the modern
world, with its abundance of text, people often want to
go even quicker. Is there a special form of silent reading
in which speed and accuracy are both high? Even better,
is there a form that we can use effectively without having
to spend time on training and practice? We turn now to a
review of research on reading via RSVP, in which words
are presented rapidly one at a time. This presentation
technique is used in some speed-reading technologies
that claim to be able to produce rapid reading with preserved comprehension and without the need for much, if
any, practice. It is important, therefore, to find out what
research shows about reading via RSVP.
Rapid Serial Visual Presentation
Presenting text with the RSVP method allows a reader to
maintain fixation at a single location on a display as a
sequence of words is presented in quick succession. The
execution of eye movements is not required at all. The
RSVP method was originally used as a tool to scientifically study reading-comprehension processes (e.g.,
Forster, 1970; Forster & Ryder, 1971; Holmes & Forster,
1972; Potter, 1984). As we will see, subsequent investigations have provided a rich body of information about
how reading is altered by this special method, information that is important for evaluating speed-reading technologies that use it.
Reading single sentences with RSVP
In research using RSVP, a short series of words is presented, with each word in view for about 100 ms (i.e., at
a speed of about 600 wpm; this duration varies across
studies). Then the participant is asked to recall the words.
Strikingly, when the words form a meaningful sentence,
participants are quite good at this task. When the words
are unrelated or scrambled, however, performance is
poor (Potter, Kroll, & Harris, 1980). The fact that a reader
can easily understand and remember a single RSVP sentence shows that words can be recognized and the meaning of a sentence can be understood very rapidly. Success
with a single sentence might be attributed to readers’
holding words in memory until the end of the sentence
and then assembling them into meaningful propositions
after the rapid presentation has ended (Mitchell, 1979,
1984). That is, the words may have been presented at a
faster rate than the comprehension system could follow.
However, there is evidence that sentence comprehension
can occur even during the course of RSVP presentation
of single sentences. For example, Masson (1986) presented sentences via RSVP at 100 ms per word (equivalent to 10 words per second or 600 wpm). The sentences
were constructed so that the final word appeared in
uppercase letters and remained on the screen until the
reader made a response to it (in one experiment, readers
made a lexical decision; in another, they named the word
aloud). For some sentences, the final word was highly
predictable from the preceding sentence context (e.g.,
“He mailed the letter without a STAMP”), and in other
How Do We Read, and Can Speed Reading Help?
sentences, the final word was unrelated to the meaning
of the sentence (e.g., “He mailed the letter without a
DRILL”). Masson found that readers were able to identify
the final word more quickly when the sentence context
made it predictable, both when the readers were making
lexical decisions about the words and when they were
naming them aloud. Importantly, the effects of predictability were substantial for both tasks, although somewhat less than those reported in earlier studies in which,
instead of RSVP, people read the sentences at a normal
rate and then made the lexical decision or naming
response (Fischler & Bloom, 1979; Stanovich & West,
1983).
To show that the RSVP predictability effects were truly
due to successful sentence comprehension, Masson
included a condition in his study in which sentence contexts were scrambled (e.g., “the without letter a mailed He
STAMP”). Predictability effects were much weaker for the
scrambled sentences than for the normal sentences, indicating that sentence comprehension contributed to those
effects. Overall, these results showed that participants
could understand RSVP sentences as they were presented
at the rate of 10 words per second, even when the task
required them to attend only to the final uppercase word.
Masson’s (1986) study showed effects of sentence context on reading of the final word of an RSVP sentence, but
such effects can also be observed for words that appear in
the middle of the sentence. In a study by Potter, Moryadas,
Abrams, and Noel (1993; see also Potter, Stiefbold, &
Moryadas, 1998), participants read sentences such as “The
sailor washed the dack of that vessel” presented via RSVP
at 10 words per second. Each sentence included either a
nonword (“dack”, which is meaningless in English) or a
real word (“duck”, which made no sense in the sentence)
that looked similar to an appropriate word (“deck”).
Readers were instructed to view the RSVP stream and
then write down exactly what they had seen, including
unexpected words and misspelled words. Despite this
warning, people misreported the nonword (“dack”) as
“deck” 40% of the time and the inappropriate word
(“duck”) as “deck” 26% of the time. These and other similar results indicate that a largely unconscious process
combines letter-level information and the sentence context in order to identify words, with a bias toward actual
words and toward words that are plausible in context.
The effect of context on these errors suggests that readers
were processing the meaning of the sentence even though
they were reading at more than twice the normal rate.
Reading more than single sentences
with RSVP
Although RSVP readers can successfully read individual
sentences, this does not necessarily mean that they can
21
comprehend longer texts. With something like a newspaper article or a book chapter, readers must not only
understand individual sentences but also build an ongoing mental model of the entire text. Do RSVP readers
comprehend and remember what they have read as well
as readers of normal text?
Two studies of text comprehension using RSVP indicated that this method led to comprehension equivalent
to that of normal reading when total presentation time
was equal to that of normal reading. When the RSVP
stream was sped up, however, comprehension and memory suffered. In one study, participants read short paragraphs that were difficult to make sense of without a title
and then recalled them (Potter et al., 1980). One such
paragraph (adapted from Bransford & Johnson, 1972)
was as follows:
The procedure is quite simple. First you arrange
everything into different groups. One pile may be
enough if you don’t have much to do. Then you
have to go somewhere else if you do not have a
machine. You put them into the machine and turn
it on. It is better not to put too many in at once.
Then you sit and wait. You have to stay there in
case anything goes wrong. Then you put everything
in another machine and watch it go around. When
it stops, you take the things home and arrange them
again. Then they can be put away in their usual
places. Soon they will all be used again and you
have to do it all over. [Doing your laundry/The
whole thing] can be a pain.
One group of participants read these paragraphs via
RSVP at one of three rates equivalent to 200, 400, or 600
wpm; another group read the paragraphs on printed
pages for equivalent durations of time per paragraph.
The key information was presented at the beginning,
middle, or end of the paragraph or was omitted (in this
example, both the phrase presenting key information
and the alternative phrase with that information omitted
are shown in brackets). All readers had a much easier
time understanding and recalling the passage when the
topic was mentioned in the first sentence than when it
was mentioned only in the last sentence or not at all,
showing that they were using the information to comprehend the subsequent sentences. At all rates, including
600 wpm, over 80% of the RSVP readers recalled the key
topic word (laundry, in this example) when it was present anywhere in the paragraph, but few guessed it when
it was omitted. Nonetheless, detailed recall accuracy
dropped markedly as the rate of presentation increased.
For participants who read the paragraphs in normal
printed form, recall was high for the first half of the paragraph but dropped markedly for the second half when
Rayner et al.
22
the allotted reading duration was equivalent to that for
the higher RSVP rates, because participants could not finish reading the entire paragraph. These readers recalled
the topic word when it was presented in the first sentence but often omitted it when it was presented in the
middle or at the end. Thus, while average accuracy was
similar for normal reading and RSVP when the text was
presented at about 200 wpm (a normal to slow reading
rate for skilled readers), what the readers were able to
remember was somewhat different between the two
forms of presentation: At higher rates, a crucial word was
more likely to be missed by normal readers than RSVP
readers.
Juola, Ward, and McNamara (1982) carried out a similar study and found similar comprehension scores
between normal reading and RSVP when overall wpm
was equated. However, as in Potter et al.’s (1980) study,
performance dropped by about 50% when the rate was
increased, either by speeding up the RSVP stream or giving readers less time with the paragraphs. Overall, the
results suggest that if you have a limited time to read
something, RSVP may be a good way to see the text without missing anything. However, the time saved will be at
the cost of relatively poor comprehension and memory.
In both of the studies we have been discussing ( Juola
et al., 1982; Potter et al., 1980), a brief pause was inserted
between each sentence. This would have allowed comprehension processes to catch up to some extent with
the rapid presentation of the text stream. As we mentioned earlier, studies of eye movements in normal reading show that readers typically pause briefly at the end of
sentences. Masson (1983) showed that without these
pauses, comprehension success was reduced in the RSVP
condition relative to a condition in which people
skimmed passages presented in full view for the same
duration.
In summary, although immediate comprehension may
be successful with single sentences presented using RSVP
speeds well beyond typical reading rates, scaling up to
full text passages yields substantial comprehension costs.
As discussed earlier, comprehension is usually assessed
through the ability to recall or answer questions about
text, which relies on the text’s being encoded into longterm memory. This encoding appears to be particularly
compromised by speed-reading procedures, including
RSVP (Masson, 1986).
Increasing Reading Speed and
Maintaining Comprehension: Proposals
and Evaluations
Now that we have reviewed the scientific evidence on
normal reading and reading with RSVP, we are in a position to look in detail at the claims of speech-reading
courses and technologies and to consider whether these
claims are valid. In the sections that follow, we ask
whether these methods can fulfill the promise to dramatically increase reading speed without hurting comprehension. In addition, we review other potential methods to
help readers get through an enormous amount of text
efficiently: effective skimming and practice.
Speed-reading courses
Speed reading was launched into public popularity in the
United States with the introduction of the Evelyn Wood
Reading Dynamics program in 1959. Evelyn Wood was a
high school teacher who claimed that she could read
very fast with good comprehension by grasping whole
phrases in a single glance. Together with her husband,
she formed a company to teach speed reading via this
method. Very quickly the number of franchises throughout the United States grew. Wood and her husband eventually sold the company, but the basic principles that she
advocated still form the foundations of most current
speed-reading training courses. These courses attempt to
teach students to change certain aspects of the reading
process while maintaining the same input method, such
as a book or computer display.
Increasing the perceptual span. Advocates of training courses make a number of assumptions about how
silent reading occurs and how speed can be increased
without sacrificing comprehension. Central to these
claims is the idea that our brain is rather lazy and that we
effectively process much less than we are actually capable of. A specific claim is that readers have the ability take
in much more information in a single glance than they
normally do and that training can “unlock” this ability.
For example, in two-thirds of a sample of 40 books on
speed-reading methods reviewed by Brozo and Johns
(1986), the claim is made that a reader’s span of recognition can be improved with practice. Most of the books
also strongly advocated against making regressive eye
movements while reading. A survey of a sample of more
recent publications that we undertook revealed that they
contained similar ideas, including recommendations to
use peripheral vision to expand the number of words
that can be read in a single fixation and to reduce regressive eye movements. The argument is that, after training,
speed readers can process entire groups of words and
phrases in a single fixation. At the extreme, it is claimed
that speed readers can zigzag down one page and up the
other page, processing the information in the text much
more efficiently than normal skilled readers do.
The evidence that we have reviewed on normal reading challenges these claims. First, what limits our ability
to process text is our capacity to recognize words and
How Do We Read, and Can Speed Reading Help?
understand text (e.g., Miellet et al., 2009). It is highly
unlikely that we can increase this ability by learning to
make eye movements differently. Second, processing
words out of order from the sensible sequence of the
sentence (Masson, 1986) or when some of the words are
removed (Fisher & Shebilske, 1985)—as would happen
when a speed reader uses a zigzag movement—impairs
the ability to process and understand the words. Third,
regressive eye movements actually support comprehension (Schotter, Tran, & Rayner, 2014) rather than causing
a problem for reading. Together, these facts about reading suggest that the ability to read is limited by our ability
to attend to, identify, and understand words rather than
our ability to see them. Even if our vision could be
improved, we would not necessarily read faster. This
conclusion is consistent with the finding that attempts to
train readers to use peripheral vision more effectively
with simple perceptual tasks that flash words in the
periphery have not been successful in allowing them to
read more quickly with good comprehension (Brim,
1968; Sailor & Ball, 1975).
Suppressing the inner voice. Another claim that
underlies speed-reading courses is that, through training,
speed readers can increase reading efficiency by inhibiting subvocalization. This is the speech that we often hear
in our heads when we read. This inner speech is an
abbreviated form of speech that is not heard by others
and that may not involve overt movements of the mouth
but that is, nevertheless, experienced by the reader.
Speed-reading proponents claim that this inner voice is a
habit that carries over from fact that we learn to read out
loud before we start reading silently and that inner
speech is a drag on reading speed. Many of the speedreading books we surveyed recommended the elimination of inner speech as a means for speeding
comprehension (e.g., Cole, 2009; Konstant, 2010; Sutz,
2009). Speed-reading proponents are generally not very
specific about what they mean when they suggest eliminating inner speech (according to one advocate, “at some
point you have to dispense with sound if you want to be
a speed reader”; Sutz, 2009, p. 11), but the idea seems to
be that we should be able to read via a purely visual
mode and that speech processes will slow us down.
However, research on normal reading challenges this
claim that the use of inner speech in silent reading is a
bad habit. As we discussed earlier, there is evidence that
inner speech plays an important role in word identification and comprehension during silent reading (see
Leinenger, 2014). Attempts to eliminate inner speech
have been shown to result in impairments in comprehension when texts are reasonably difficult and require readers to make inferences (Daneman & Newson, 1992;
Hardyck & Petrinovich, 1970; Slowiaczek & Clifton, 1980).
23
Even people reading sentences via RSVP at 720 wpm
appear to generate sound-based representations of the
words (Petrick, 1981).
Evaluations of trained speed readers. Although the
results of the studies we have reviewed suggest that the
claims of speed-reading courses are overstated, it could be
argued that data from people who read at typical speeds
cannot be generalized to speed readers. Perhaps speed
readers are doing something very different from typical
readers. Indeed, studies of extraordinary performers in
other cognitive domains have hinted at this idea. For
example, the average person is capable of remembering
only about 7 items (e.g., random digits, letters, words) in
their presented order; larger sequences lead to various
types of memory errors. However, several years ago, two
researchers followed an undergraduate whose initial performance was average but climbed to 80 items after 20
months of practice (Ericsson, Chase, & Faloon, 1980). As a
long-distance runner, he was familiar with the format of
running times (i.e., 3- or 4-digit sequences), so he chunked
the digits into this format and built those chunks into a
hierarchical structure for storage in long-term memory.
Rather than expanding his short-term memory capacity, he
changed the task to use long-term memory, a system with
a vast capacity. His increase in performance did not generalize, however: His performance on the task with letters
instead of digits fell back to normal levels. Thus, although
he learned some special skills (i.e., his new strategy), he
did not overcome the cognitive limitations that constrain
others (i.e., short-term memory capacity). Is something
similar going on with trained speed readers? There is not a
great deal of well-controlled research on this topic, but we
turn now to some studies that have examined speed readers and have evaluated their eye movements, reading
speed, and comprehension.
Llewellyn-Thomas (1962) and McLaughlin (1969) each
recorded the eye movements of one speed reader and
found that, as the training courses recommend, the speed
readers moved their eyes down the middle of the lefthand page and then up the middle of the right-hand
page, fixating particular lines only once and completely
skipping most lines altogether. Using this peculiar pattern
of eye fixations, the reader processes half of the material
(i.e., the right-hand pages) in the sequence opposite to
that in which it was written (and is normally read), and
therefore in a different order than the author intended. As
you might expect, when McLaughlin (1969) tested comprehension with free recall, he found that the speed
reader recalled confused—and sometimes completely
fabricated—information from the text. The poor comprehension calls to mind the comic Woody Allen’s classic
line: “I took a speed reading course where you run your
finger down the middle of the page and was able to read
24
War and Peace in 20 minutes. It's about Russia.” ('War
and Peace' in 20 Minutes?, 1995).
Another study looked at two graduates of a speedreading program who were considered by program officials to have achieved such remarkable performance that
the officials contacted a cognitive scientist to test them
under controlled conditions in a laboratory (Homa, 1983).
When so tested, each student read an entire college-level
textbook in less than 6 minutes, achieving rates of 15,000
wpm or higher. Although the students speed read the
book three times, their performance on a multiple-choice
test of comprehension was quite poor. Neither speed
reader showed any extraordinary ability, as compared to
average readers, in perceiving peripherally presented letters or identifying which words had appeared in a briefly
presented paragraph. The extraordinary ability that they
had achieved, the investigator concluded, was “a remarkable dexterity in page-turning” (Homa, 1983, p. 126).
Calef, Pieper, and Coffey (1999) recorded the eye
movements of a group of adults both before and after
they enrolled in a speed-reading class, comparing them
to a group of people who did not take the class. In the
pretest, both groups of readers read at about 280 wpm.
After the speed-reading course, the speed readers read at
about 400 wpm, making fewer fixations (and regressions)
with shorter fixation durations (228 ms after compared
with 241 ms before). Their comprehension score
decreased from pretest (81% correct) to posttest (74%),
indicating that the increased rate of speed was achieved
at the expense of comprehension.
Further evidence for comprehension difficulties among
speed readers came from a study by Liddle (1965; reanalyzed by Carver, 1971, 1972), who tested graduates of
Wood’s speed-reading program and compared them to
readers who had signed up for the program but had not
yet taken the course. Both groups of people were tested
for both speed and comprehension on fiction and nonfiction material. The reading rates were about 300 to 1,300
wpm faster for the graduates than the control group.
While test scores revealed a significant decline in comprehension among the graduates on the fiction material,
the two groups showed approximately the same level of
comprehension for the nonfiction material (68% for the
graduates and 72% for those who had not yet taken the
course). But in nonfiction material, the content is based
on the real world. Thus, it is possible that the speed readers could have answered the questions correctly by
knowing the answers rather than actually having read
them from the text. In fact, when Carver administered the
same comprehension test to a group of people who had
never seen the passage, they obtained an only slightly
lower score (57%) through their use of common knowledge and guessing. This qualifies the conclusion that
there was no comprehension loss for the nonfiction
Rayner et al.
material (in addition to the already established decrease
in comprehension for fiction material).
The most complete study of the eye movements and
comprehension of speed readers was carried out by
Just, Masson, and Carpenter (1980; see also Just &
Carpenter, 1987). They presented passages to speed
readers (reading rates around 600–700 wpm), normal
readers (reading rates around 250 wpm), and people
who were asked to skim (producing rates around 600–
700 wpm). The speed readers did better than skimmers
on general comprehension questions about the gist of
the passages but not quite as well as people reading at
normal speed. Normal readers, who made many more
fixations than the speed readers, were able to answer
questions about details of the text relatively well, while
skimmers and speed readers, who made many fewer
fixations than normal readers, did not differ from each
other on these items. They could not answer these questions if they had not fixated on the regions where the
answers were located. The data thus suggest that the
students of speed-reading courses are essentially being
taught to skim and not really read in the sense that we
use the term “reading” here. The advantage of trained
speed readers over skimmers with respect to general
comprehension of the text was ascribed by Just and colleagues to an improvement in what they called extended
inferencing. Essentially, the speed readers had increased
their ability to construct reasonably accurate inferences
about text content on the basis of partial information
and their preexisting knowledge. In fact, when the three
groups of participants were given more technical texts
(taken from Scientific American), for which background
knowledge would be very sparse, the speed readers no
longer showed an advantage over the skimmers, even
on general questions.
So what about phenomenally fast speed readers such
as Anne Jones and Howard Berg, mentioned at the beginning of this article? One advantage Jones had in reading
the new Harry Potter book was having read earlier books
in the series. That experience probably allowed her to
capitalize on a large amount of background knowledge
about such things as characters, plot structure, and writing style. Combining that background knowledge with
visual sampling from pages of the new book and a highly
developed ability to engage in extended inferencing ( Just
et al., 1980) could have allowed her to generate a coherent synopsis of the book. A more thorough assessment of
comprehension achieved by speed readers like Jones and
Berg, based on tests such as those used in some of the
research studies we have reviewed, is lacking. Moreover,
as we have seen, it is very important to assess how well
a reader can perform on a comprehension test based
merely on background knowledge, without having read
the critical text. So, at this point, we can only speculate
How Do We Read, and Can Speed Reading Help?
about what an objective assessment of the reading comprehension of Jones and Berg would reveal.
To summarize, there is no evidence that training programs allow people to dramatically increase their reading
rates while maintaining excellent comprehension. If
speed readers have developed a special skill beyond the
ability to turn pages quickly, that skill may be learning to
skim. Effective speed readers appear to be intelligent
people who already know a great deal concerning the
topic they are reading about and are able to successfully
skim the material at rapid rates and accept the lowered
comprehension that accompanies skimming (Carver,
1985). This brings us to a discussion of skimming as a
strategy to deal with the problem of having too much to
read and too little time.
Skimming
Although we have shown that many elements of speedreading training are not likely to be effective, some elements may be used adaptively to more efficiently process
text. Reading faster necessarily means not reading a text
in its entirety, which essentially means skimming. Taylor
notes that Wood “repeatedly stated that her people are
not skimming, but rather are reading” (Taylor, 1962,
p. 65). Based on recordings of their eye movements,
however, Taylor concluded that they closely resembled
the eye movement patterns produced during skimming
(Taylor, 1965; see also Walton, 1957). He tested a large
group of graduates from Wood’s training program and
found that those who more often moved down the center
of the page had the poorest comprehension of the text
on a true/false test (around 50%—chance performance).
So, if speed reading essentially boils down to skimming,
are there more effective ways to skim?
Research on skimming has revealed important facts
about what it entails and how it affects comprehension.
One goal of skimming a text is to obtain a general idea
about its content, extracting the important information.
Therefore, the suggestion from speed-reading proponents that speed readers should zigzag down one page
and up the other may be the least effective way to skim.
More deliberate skimming, with more time spent on the
critical portions of the text, may be more effective. In a
study comparing people reading normally and skimming
at about twice their normal reading rate, Masson (1982)
showed that people who were skimming were less able
to correctly identify statements taken from the text,
regardless of whether the statements reflected important
information or minor details. The equal drop in performance for important and for detailed information indicated that skimmers were not able to effectively select
important information to read carefully while skipping
minor details (see also Carver, 1984; Dyson & Haselgrove,
2000). Perhaps, then, the skimmers who were tested in
25
these studies were not using effective strategies to do so.
Are there more effective ways?
Effective skimmers scan a text for headings, paragraph structure, or key words to locate potentially relevant information, then read more carefully when such
regions are found. Research has shown that readers
who pay more attention to headings write the most
accurate text summaries (Hyönä, Lorch, & Kaakinen,
2002). Indeed, a number of recent speed-reading books
have provided helpful advice along these lines. For
example, Konstant (2010) advocated inspection of a
book’s structure (e.g., its table of contents) and selective
reading of the first paragraph of each section and the
first sentence of every paragraph to help find important
information. The central idea is that one does not need
to read the same way for every reading goal. If it is necessary not to understand all aspects of a text but rather
to find a particular fact, then selective reading such as
skimming can help people to achieve their goal more
efficiently. Duggan and Payne (2009) showed that skimming an entire text led to better memory for important
ideas than normally reading only the first half or only
the second half of the text, but skimming comprehension was equivalent to when subjects read only half of
every paragraph at normal speed. The eye movements
revealed that skimmers tended to spend more time
reading earlier paragraphs and earlier pages, suggesting
that they used the initial parts of the text to obtain the
general topic of passages and provide context for the
later parts that they skimmed in a more cursory way.
Therefore, effective skimming means making sensible
decisions about which parts of a text to select for more
careful reading when faced with time pressure. In fact,
Wilkinson, Reader, and Payne (2012) found that, when
forced to skim, readers tended to select texts that were
less demanding, presumably because they would be
able to derive more information from such texts when
skimming. This kind of information foraging is a useful
method of handling large amounts of text in a timely
manner.
To conclude, skimming is an important skill and may
be a reasonable way to cope with the overwhelming
amount of text we have to read, as long as we are willing
to accept the trade-off between speed and accuracy that
skimming requires. Strategies such as attending to headings and spending more time on the beginning and ending of paragraphs may improve comprehension during
skimming or may allow people who are skimming to
access the information they seek more effectively.
Speed-reading technologies
Although speed-reading training courses are still somewhat popular (most colleges have such a course somewhere on campus), their appeal has waned and has been
Rayner et al.
26
Fig. 14. A depiction of a rapid-serial-visual-presentation (RSVP) display as would be used on a mobile
device. Words are presented one at a time on the screen at a device-set speed.
replaced by the appeal of speed-reading technologies.
Many of these technologies are based on the claim that
the eye movements we make during reading are a waste
of time and that we would be much more efficient if we
used a device to present words to us at a rate faster than
we would normally receive them via eye movements. A
further claim is that the reader need not have control
over the exact sequence and timing with which the words
are presented; the device can take control of these things,
leaving the reader more time to process the words and
their meanings.
RSVP-based technologies. Recently popular methods
present words one at a time on a digital display using
the RSVP technique that we discussed earlier. The
appeal of such methods may reflect the fact that advocates of many of these technologies claim that no training in use of the technology is necessary. The idea is
that the computer or smartphone does all the work,
presenting words for the reader in rapid succession. The
person can simply sit back and passively absorb the
text. Because the device controls the timing, users can
set their reading rate to whatever they want (e.g., 200,
500, or 1,000 wpm), regardless of what their comfortable, natural reading rate is (Fig. 14).
One recent method of this kind is used by apps based
on a technology called Spritz. There are several competitor apps that are based on the same principle, some of
which preceded the development of Spritz, but because
Spritz has received the most attention, and because its
developers have made many specific claims about the
science behind it, we spend the majority of this section
addressing those claims directly. For instance, Spritz
included the following passage under the heading
“Reading Basics” on its website, which we will walk
through part by part below:
Traditional reading involves publishing text in lines
and moving your eyes sequentially from word to
word. For each word, the eye seeks a certain point
within the word, which we call the “Optimal
Recognition Point” or ORP. . . Once the ORP is
found, processing the word for meaning and
context occurs and your eyes move to the next
word. When your eyes encounter punctuation
within and between sentences, your brain is
prompted to assemble all of the words that you
have read and processes them into a coherent
thought. (The Science, 2015)
Consider, first, the website’s mention of the optimal
recognition point, which should sound familiar if you
recall the optimal viewing position, or OVP (O’Regan et
al., 1984), described earlier. However, contrary to the
implication of the Spritz website, readers are capable of
identifying words whether or not they look at the OVP;
there is just a slight loss of processing efficiency if their
eyes land on a different point in the word. Moreover, the
research we reviewed on parafoveal preview benefit suggests that readers begin to process words before looking
directly at them, rather than “once the ORP is found.” To
be fair to Spritz, centering words at the OVP may be an
improvement over RSVP methods in which words are left
justified. This is because, with left justification, words of
different lengths will be placed such that their OVPs are
in different locations and the reader will probably be
viewing the word from a non-optimal position. However,
when words are centered, as in the RSVP paragraph
experiment described earlier (Potter et al., 1980), the eye
will remain close to the OVP.
The idea that the brain assembles the words and processes them into a coherent meaning only when punctuation marks are encountered is also problematic. As we
How Do We Read, and Can Speed Reading Help?
have discussed, research shows that people identify words,
sentence structures, and meanings incrementally. Even
though some additional wrap-up processes occur at the
end of a sentence (Just & Carpenter, 1980; Warren et al.,
2009), people do not wait until the end of a phrase or
sentence to compute its meaning (Frazier & Rayner, 1987;
von der Malsburg & Vasishth, 2013).
Spritz has also been quoted by media outlets as saying
that only 20% of the reader’s time is spent processing
content while 80% of the time is spent moving the eyes
(e.g., Luchette, 2014). First, let us address the implication
that the time spent processing content and moving the
eyes must be mutually exclusive. For this to be true, all
cognitive processing must halt during an eye movement,
which is not the case (Irwin, 1998). Instead, the brain
continues processing the information it had obtained on
the prior fixation. Moreover, saccades during reading last
approximately 20 to 35 ms (we will use 25 ms for simplicity), and fixations last approximately 250 ms. Thus, our
estimate of the percentage of time spent moving the eyes
(between one fixation and the next) is only about 10% of
the total reading time. If we assume that we are always
processing content while reading (e.g., recalling prior
parts of the text, identifying the current word, making
sense of the sentence as a whole), then a more accurate
statement might be that 100% of the time is spent processing content and, during that time, the eyes are moving 10% of the time.
In a blog post on their website, Spritz also claimed
that “every saccade has a penalty in both time and
comprehension” and that “right-to-left saccades are discombobulating for many people” (Why Spritz Works,
2014). However, as we discussed earlier, eye movements
are under control of the brain and are used to access the
text in precisely the way we need it to be presented.
Thus, they certainly do not waste time. Eye movements
are a useful aspect of the reading process because they
are sensitive to, and can compensate for, difficulties
understanding the text. They do so by dwelling longer on
difficult or unexpected words and by permitting regressions (Schotter, Tran, & Rayner, 2014). Readers are often
unaware of the regressions they make, unless they deliberately pay attention to what their eyes are doing. A
regression is more likely to be the solution to discombobulation (i.e., rereading to fix a failure in comprehension) than a cause of it.
One possible solution to the inability to reread in
RSVP would be to incorporate a “go back” button on the
device that would repeat the text for the reader, a feature
that many apps using RSVP technology, including Spritz,
have added. However, many of these buttons merely start
the text (or sentence) over from the beginning. One
recently announced version of RSVP technology to be
released for the Amazon Kindle claims that it will allow
27
readers to use their finger to scan through the text continuously in order to move backward or forward to a
particular word. This is an improvement over only being
able to start at the beginning of a sentence in that it
allows for more flexibility in moving through the text.
However, it has not been shown that these methods for
regressing (or skipping forward) in RSVP are as effective
as the eyes are in conventional reading, in which readers
have a spatial representation of the text and a sense of
where to move their eyes to clarify something and where
to return to continue reading. In order to do this in a truly
effective way, the device would need to know how far
back in the text to go and would need to repeat the text
from that point. Regression behavior in reading is one of
the less understood aspects of the process, so even generating a general principle for this would be difficult—
not to mention that there is substantial variability both
across and within individuals in the sequence with which
they reread words (von der Malsburg & Vasishth, 2011).
Another concern about apps using RSVP is that they
do not allow for parafoveal preview. As we have discussed, a large body of research using the moving-window paradigm (see Rayner, 2014) and boundary paradigm
(see Schotter et al., 2012) has shown that information
from more than just the currently fixated word is used.
Some information obtained from the next word before it
is directly fixated is used to give the reader a head start
on processing it; if this preview is denied (e.g., if the
word is masked until it is fixated or not presented at all),
reading time for the word is longer than if it had been
visible. With the single-word RSVP method, readers do
not have, and therefore cannot take advantage of, preview information.
An important concern about RSVP-based apps regards
the limits that such a technique may pose for comprehension. In the “Rapid Serial Visual Presentation” section
above, we described a number of studies that showed
similar levels of comprehension for full text reading and
RSVP when words were presented at a normal rate. It
appears that, at least when comprehension is measured
as it was in those studies, the elimination of parafoveal
preview and of opportunities for regressive movements
to earlier parts of a text do not markedly undermine text
understanding when RSVP is used. But two crucial caveats must be emphasized. First, when pauses between
sentences were not included in RSVP presentations of
text, comprehension clearly suffered (Masson, 1983).
Second, speeding text presentation beyond normal reading rates consistently led to reduced levels of comprehension and memory ( Juola et al., 1982; Potter et al.,
1980). These results indicate that the mental operations
responsible for assembling viewed words into meaningful ideas and retaining them in memory cannot be completed if adequate time is not provided. Therefore, the
Rayner et al.
28
maintaining good comprehension. The real practical application of RSVP may be for wearable technology like smartwatches and glasses with optical displays that have only
enough space to present one or two words at a time. The
RSVP method may be the only viable way to display text on
these screens. If the message is not too long or complex,
RSVP may suffice to get it across.
Fig. 15. A depiction of the BeeLine Reader display as presented on the BeeLine Reader website (www.beelinereader.com). Words in the sentences are
colored depending on their position in a line. Text transitions along each
line in a gradient (red, blue, black, and/or shades of gray) such that the end
of one line and the beginning of the next are presented in the same color.
Reprinted from the BeeLine Reader website (http://www.beelinereader
.com/). Copyright 2015 by BeeLine Reader. Reprinted with permission.
promise that RSVP can produce faster reading without
compromising understanding and memory is not supported by the research we reviewed.
As of the writing of this article, we are familiar with
only two published studies that directly compared traditional reading to reading with (an approximation of)
Spritz (Benedetto et al., 2015; Dingler, Shirazi, Kunze, &
Schmidt, 2015). Only one of these studies objectively
compared comprehension in RSVP reading and con­
ventional reading (Benedetto et al., 2015). That study
involved a passage of text taken from George Orwell’s
1984, and participants were allowed to pause the RSVP
stream whenever they wanted. Taking these pauses into
account, the average reading speed in the traditional
reading condition and that in the RSVP condition were
very similar and somewhat slower than normal reading
(200 and 209 wpm, respectively). Reading via RSVP
increased eye fatigue relative to traditional reading and,
although accuracy was equivalent on comprehension
questions that required making inferences, it was poorer
on comprehension questions that assessed literal comprehension of the text. These data suggest that readers
were still able to get the gist of the passage when reading at normal rates via RSVP but that their ability to
maintain a veridical representation of the wording of
the text was impaired. It is likely that even the gist representation would have been impaired at faster RSVP
rates. Additional work needs to be done to evaluate
reading using technologies such as Spritz, and this work
will need to be presented in peer-reviewed conferences
and journals so that it can be evaluated by the scientific
community.
Based on the research that has been done on normal
reading and RSVP, we suspect that presentation of
moderately complex texts via RSVP apps will not allow
people to achieve the goal of greatly increasing reading
speed (e.g., by doubling or tripling their normal rate) while
Color-based technology. Another technology that has
been suggested as a way of making reading faster and
easier but that has not received as much attention as
RSVP-based procedures relies on manipulating the coloring of a text. The BeeLine Reader (Fig. 15) colors lines of
text in a gradient from one end to the other, such that the
end of one line and the beginning of the next are colored
similarly. The idea behind this approach is that a significant problem in reading is the ability to make accurate
return sweeps and that reducing these errors will improve
reading. As we have discussed, however, return sweeps
are not the complicated part of reading ( Just & Carpenter, 1980)—language processing is. Thus, any savings in
reading time provided by making return sweeps more
efficient may constitute only a minimal improvement in
the efficacy of reading as a whole.
Practice reading
As we have suggested, effective skimming can be one
way to increase reading speed while maintaining at least
a moderate level of comprehension and to ascertain a
certain keyword or fact or the gist of a topic. What other
research-tested methods are there for increasing speed
without sacrificing comprehension? As we will see in this
section, modest improvements are possible, but there are
cognitive and visual limitations that cannot be ignored.
Practice with reading, and in particular reading to
comprehend, does help. But it helps slowly and does
not cause drastic increases in speed. The importance of
practice is widely appreciated for skills such as playing
sports or musical instruments. We would not throw
someone into a pool and expect them to swim without
having learned and practiced the skill of swimming. We
would not expect someone to lift weights over and over
and become a better football player, because practice
means engaging in a complex task in its entirety. Practice
is also important for cognitive skills, and reading is no
exception. The kind of practice that will help reading is
practice that helps people to identify words and comprehend better, not just take in visual information faster.
Indeed, as we saw earlier, reading speed is related more
to one’s language-processing skills than to the ability to
control one’s eyes. If visual information comes in faster
than the comprehension system can process it, the
increased speed is wasted on broken comprehension.
How Do We Read, and Can Speed Reading Help?
The practice that is required to become a better reader is
thus practice with language—a conclusion that is not
surprising, given that language is what writing represents. Moreover, written language uses some vocabulary
and syntactic structures that are not commonly found in
speech, and practice with reading can give people practice with these.
One specific example of how practice with language
can help reading is related to the effects of word frequency that we discussed earlier. As we saw, fixation
durations are shorter for common words such as house
than for uncommon words such as abode (e.g., Rayner &
Duffy, 1986; see Rayner, 1998, 2009). However, as the
reader sees abode more and more, its frequency in the
reader’s experience increases. Thus, the more often you
read an uncommon word, the more common it becomes,
and you will be able to read that word more easily and
quickly in the future (Acheson, Wells, & MacDonald,
2008). In addition to making individual words more frequent, reading can introduce you to new words, increasing their frequency above zero for you. Reading more,
and reading varied texts, will increase your vocabulary
and will benefit reading more than reading the same passage again and again.
Another example of how practice with language can
help reading comes from the findings reviewed earlier
showing that context has a strong influence on fixation
durations (Ehrlich & Rayner, 1981; Zola, 1984; see
Rayner, 1998, 2009). That is, readers spend less time on
words that are more predictable because of the prior
sentence context than words that are not predictable in
context. Therefore, more experience with language will
lead you to generate more and better expectations about
upcoming words and to be better at extended inferencing. This is why it is important to read a diverse set of
texts: Expanding your knowledge through reading about
different topics will allow you to generate better expectations about what the text is about to say. That may be
the basis for some anecdotes about the speed-reading
abilities of famous people, such as that President
Kennedy could pick up a copy of the Washington Post
or the New York Times and read it from front to back in
a few minutes. However, consider the knowledge and
information that someone like Kennedy would bring to
the task of reading the newspaper. As president, he was
briefed about important world events each day and was
involved in generating much of the policy and events
reported in the newspaper; thus, he probably had firsthand knowledge of much of what was described. In
contrast, the average person would come to such a situation with very few facts at his or her disposal and would
probably have to read an article rather carefully in order
to completely understand it. To read rapidly, you need to
know enough about a topic to fit the new information
29
immediately into what you already know and to make
inferences.
Conclusions
Many people wish to read faster by finding a special form
of reading in which they read more quickly with excellent
comprehension, ideally without much effort or training. In
this article, we have seen that there is no such magic bullet. There is a trade-off between speed and accuracy in
reading, as there is in all forms of behavior. Increasing the
speed with which you encounter words, therefore, has
consequences for how well you understand and remember the text. In some scenarios, it is tolerable and even
advisable to accept a decrease in comprehension in
exchange for an increase in speed. This may occur, for
example, if you already know a lot about the material and
you are skimming through it to seek a specific piece of
information. In many other situations, however, it will be
necessary to slow down to a normal pace in order to
achieve good comprehension. Moreover, you may need to
reread parts of the text to ensure a proper understanding
of what was written. Bear in mind, however, that a normal
pace for most readers is 200 to 400 wpm. This is faster than
we normally gain information through listening, and pretty
good for most purposes.
Authors’ Note
The first draft of this article was submitted a few days before
Keith Rayner passed away after a long and hard-fought battle
with cancer. The numerous studies of his described within this
article represent only a fraction of the important, influential, and
high-quality work he conducted over four decades in the fields
of reading research and cognitive psychology. Rayner’s contribution to science is profound (for a more complete tribute, see
Clifton et al., 2016), and he proposed the current article because
he felt that it was important to share the knowledge we have
gained from experimental science with the general public.
Acknowledgments
Figures 4, 5, 7, 9, 10, 11, and 13 were created for this article by
Trevor John Polcyn.
Declaration of Conflicting Interests
The authors declared that they had no conflicts of interest with
respect to their authorship or the publication of this article.
Funding
During the preparation of this article, K. Rayner and E. R.
Schotter were supported by the Atkinson Family Endowment,
National Institutes of Health Grant R01 HD065829, and a gift
from the Microsoft Corporation. M. E. J. Masson was supported
Discovery Grant 7910 from the Natural Sciences and Engineering
Research Council of Canada.
30
References
Acheson, D. J., Wells, J. B., & MacDonald, M. C. (2008). New
and updated tests of print exposure and reading abilities in
college students. Behavior Research Methods, 40, 278–289.
Angele, B., & Rayner, K. (2011). Parafoveal processing of
word n+2 during reading: Do the preceding words matter? Journal of Experimental Psychology: Human Perception
and Performance, 37, 1210–1220.
Angele, B., & Rayner, K. (2013). Processing the in the parafovea:
Are articles skipped automatically? Journal of Experimental
Psychology: Learning, Memory, and Cognition, 39, 649–662.
Apel, J. K., Henderson, J. M., & Ferreira, F. (2012). Targeting
regressions: Do readers pay attention to the left?
Psychonomic Bulletin & Review, 19, 1108–1113.
Balota, D. A., & Rayner, K. (1991). Word recognition processes
in foveal and parafoveal vision: The range of influence of
lexical variables. In D. Besner & G. W. Humphreys (Eds.),
Basic processes in reading: Visual word recognition (pp.
198–232). Hillsdale, NJ: Erlbaum.
Becker, W., & Jürgens, R. (1979). An analysis of the saccadic
system by means of double step stimuli. Vision Research,
199, 1967–1983.
Benedetto, S., Carbone, A., Pedrotti, M., Le Fevre, K., Bey, L. A.
Y., & Baccino, T. (2015). Rapid serial visual presentation in
reading: The case of Spritz. Computers in Human Behavior,
45, 352–358.
Berry, C. (1971). Advanced frequency information and verbal
responses times. Psychonomic Science, 23, 151–152.
Bouma, H. (1973). Visual interference in the parafoveal recognition of initial and final letters of words. Vision Research,
13, 767–782.
Bouma, H. (1978). Visual search and reading: Eye movements
and functional visual field. In J. Requin (Ed.), Attention and
performance (Vol. VII, pp. 115–147). Hillsdale, NJ: Erlbaum.
Bransford, J. D., & Johnson, M. K. (1972). Contextual prerequisites for understanding: Some investigations of comprehension and recall. Journal of Verbal Learning and Verbal
Behavior, 11, 717–726.
Brim, B. J. (1968). Impact of a reading improvement program.
Journal of Educational Research, 62, 177–182.
Brozo, W. G., & Johns, J. L. (1986). A content analysis of 40
speed reading books. Journal of Reading, 30, 242–247.
Calef, T., Pieper, M., & Coffey, B. (1999). Comparisons of eye
movements before and after a speed-reading course. Journal
of the American Optometric Association, 70, 171–181.
Carver, R. P. (1971). Sense and nonsense in speed reading.
Silver Spring, MD: Revrac.
Carver, R. P. (1972, August). Speed readers don’t read; they
skim. Psychology Today, pp. 22–30.
Carver, R. P. (1984). Rauding theory predictions of amount
comprehended under different purposes and speed reading conditions. Reading Research Quarterly, 19, 205–218.
Carver, R. P. (1985). How good are some of the world’s best
readers? Reading Research Quarterly, 20, 389–419.
Carver, R. P. (1990). Reading rate: A review of research and
theory. San Diego, CA: Academic Press.
Cattell, J. M. (1886). The time it takes to see and name objects.
Mind, 11, 63–65.
Rayner et al.
Clifton, C., Ferreira, F., Henderson, J. M., Inhoff, A. W.,
Liversedge, S. P., Reichle, E. D., & Schotter, E. R. (2016).
Eye movements in reading and information processing:
Keith Rayner’s 40 year legacy. Journal of Memory and
Language, 86, 1–19.
Cole, R. (2009). How to be a super reader. London, England:
Piatkus.
Coltheart, M., Rastle, K., Perry, C., Langdon, R., & Ziegler, J.
(2001). DRC: A dual route cascaded model of visual word
recognition and reading aloud. Psychological Review, 108,
204–256.
Crash Blossoms Up the Garden Path. (2009, September 24).
Sentence first: An Irishman’s blog about language [Blog post].
Retrieved from https://stancarey.wordpress.com/2009/09/
Cutter, M. G., Drieghe, D., & Liversedge, S. P. (2014).
Preview benefit in English spaced compounds. Journal
of Experimental Psychology: Learning, Memory, and
Cognition, 40, 1778–1786.
Daneman, M., & Newson, M. (1992). Assessing the importance
of subvocalization during normal silent reading. Reading
and Writing, 4, 55–77.
Dingler, T., Shirazi, A. S., Kunze, K., & Schmidt, A. (2015).
Assessment of stimuli for supporting speed reading on
electronic devices. In Proceedings of the 6th Augmented
Human International Conference (pp. 117–124). New
York, NY: Association for Computing Machinery.
Duffy, S. A., Morris, R. K., & Rayner, K. (1988). Lexical ambiguity and fixation times in reading. Journal of Memory and
Language, 27, 429–446.
Duggan, G. B., & Payne, S. J. (2009). Text skimming: The process and effectiveness of foraging through text under time
pressure. Journal of Experimental Psychology: Applied, 15,
228–242.
Dyson, M. C., & Haselgrove, M. (2000). The effects of reading speed and reading patterns on the understanding of
text read from screen. Journal of Research in Reading, 23,
210–223.
Ehrlich, S. F., & Rayner, K. (1981). Contextual effects on word
perception and eye movements during reading. Journal of
Verbal Learning and Verbal Behavior, 20, 641–655.
Engbert, R., Nuthmann, A., Richter, E. M., & Kliegl, R. (2005).
SWIFT: A dynamical model of saccade generation during
reading. Psychological Review, 112, 777–813.
Ericsson, K. A., Chase, W. G., & Faloon, S. (1980). Acquisition
of a memory skill. Science, 208, 1181–1182.
Everatt, J., Bradshaw, M. F., & Hibbard, P. B. (1998). Individual
differences in reading and eye movement control. In G.
Underwood (Ed.), Eye guidance in reading and scene perception (pp. 223–242). Oxford, England: Elsevier Science.
Everatt, J., & Underwood, G. (1994). Individual differences in
reading subprocesses: Relationships between reading ability, lexical access, and eye movement control. Language
and Speech, 37, 283–297.
Ferreira, V. S., & Schotter, E. R. (2013). Do verb bias effects on
sentence production reflect sensitivity to comprehension
or production factors? Quarterly Journal of Experimental
Psychology, 66, 1548–1571.
Fischler, I., & Bloom, P. A. (1979). Automatic and attentional
processes in the effects of sentence contexts on word
How Do We Read, and Can Speed Reading Help?
recognition. Journal of Verbal Learning and Verbal
Behavior, 18, 1–20.
Fisher, D. F., & Shebilske, W. L. (1985). There is more that
meets the eye than the eyemind assumption. In R. Groner,
G. W. McConkie, & C. Menz (Eds.), Eye movements and
human information processing (pp. 149–158). Amsterdam,
The Netherlands: North Holland.
Forster, K. I. (1970). Visual perception of rapidly presented
word sequences of varying complexity. Perception &
Psychophysics, 8, 215–221.
Forster, K. I., & Chambers, S. M. (1973). Lexical access and naming time. Journal of Verbal Learning and Verbal Behavior,
12, 627–635.
Forster, K. I., & Ryder, L. A. (1971). Perceiving the structure
and meaning of sentences. Journal of Verbal Learning and
Verbal Behavior, 10, 285–296.
Frazier, L., & Rayner, K. (1982). Making and correcting errors
during sentence comprehension: Eye movements in the
analysis of structurally ambiguous sentences. Cognitive
Psychology, 14, 178–210.
Frazier, L., & Rayner, K. (1987). Resolution of syntactic category
ambiguities: Eye movements in parsing lexically ambiguous
sentences. Journal of Memory and Language, 26, 505–526.
Friedman, R. B. (1980). Identity without form: Abstract representations of letters. Perception & Psychophysics, 28, 53–60.
Gilchrist, I. D. (2011). Saccades. In S. Liversedge, I. D. Gilchrist,
& S. Everling (Eds.), Oxford handbook of eye movements
(pp. 85–95). Oxford, England: Oxford University Press.
Goodman, K. S. (1967). Reading: A psycholinguistic guessing
game. Literacy Research and Instruction, 6, 126–135.
Gough, P. B., & Hillinger, M. L. (1980). Learning to read: An
unnatural act. Annals of Dyslexia, 30, 179–196.
Hardyck, C. D., & Petrinovich, L. F. (1970). Subvocal speech
and comprehension level as a function of the difficulty
level of reading material. Journal of Verbal Learning and
Verbal Behavior, 9, 647–652.
Hohenstein, S., & Kliegl, R. (2014). Semantic preview benefit during reading. Journal of Experimental Psychology:
Learning, Memory, and Cognition, 40, 166–190.
Holmes, V. M., & Forster, K. I. (1972). Perceptual complexity and underlying sentence structure. Journal of Verbal
Learning and Verbal Behavior, 11, 148–156.
Homa, D. (1983). An assessment of two extraordinary speedreaders. Bulletin of the Psychonomic Society, 21, 123–126.
Hyönä, J., Lorch, R. F., Jr., & Kaakinen, J. K. (2002). Individual
differences in reading to summarize expository text:
Evidence from eye fixation patterns. Journal of Educational
Psychology, 94, 44–55.
Inhoff, A. W., & Liu, W. (1998). The perceptual span and oculomotor activity during the reading of Chinese sentences.
Journal of Experimental Psychology: Human Perception
and Performance, 24, 20–34.
Irwin, D. E. (1998). Lexical processing during saccadic eye
movements. Cognitive Psychology, 36, 1–27.
Ishida, T., & Ikeda, M. (1989). Temporal properties of information extraction in reading studied by a text-mask replacement technique. Journal of the Optical Society of America
A, 6, 1624–1632.
31
Johnson, R. L. (2009). The quiet clam is quite calm: Transposedletter neighborhood effects on eye movements during reading. Journal of Experimental Psychology: Learning, Memory,
and Cognition, 35, 943–969.
Jordan, T. R., Almabruk, A. A. A., Gadalla, E. A., McGowan,
V. A., White, S. J., Abedipour, L., & Paterson, K. B. (2014).
Reading direction and the central perceptual span: Evidence
from Arabic and English. Psychonomic Bulletin & Review,
21, 505–511.
Juola, J. F., Ward, N. J., & McNamara, T. (1982). Visual search
and reading of rapid serial presentations of letter strings,
words, and text. Journal of Experimental Psychology:
General, 111, 208–227.
Just, M. A., & Carpenter, P. A. (1980). A theory of reading: From
eye fixations to comprehension. Psychological Review, 87,
329–354.
Just, M. A., & Carpenter, P. A. (1987). The psychology of reading
and language comprehension. Boston, MA: Allyn & Bacon.
Just, M., Masson, M., & Carpenter, P. (1980). The differences
between speed reading and skimming. Bulletin of the
Psychonomic Society, 16, 171.
King, J. W., & Kutas, M. (1998). Neural plasticity in the dynamics of human visual word recognition. Neuroscience Letters,
244, 61–64.
Konstant, T. (2010). Work smarter with speed reading. London,
England: Hodder Education.
Kuperman, V., & Van Dyke, J. A. (2011). Effects of individual
differences in verbal skills on eye-movement patterns during sentence reading. Journal of Memory and Language, 65,
42–73.
Kutas, M., & Federmeier, K. D. (2011). Thirty years and counting: Finding meaning in the N400 component of the eventrelated brain potential (ERP). Annual Review of Psychology,
62, 621–647.
Kutas, M., & Hillyard, S. A. (1984). Brain potentials during
reading reflect word expectancy and semantic association.
Nature, 307, 161–163.
Leinenger, M. (2014). Phonological coding during reading.
Psychological Bulletin, 140, 1534–1555.
Leinenger, M., & Rayner, K. (2013). Eye movements while reading biased homographs: Effects of prior encounter and
biasing context on reducing the subordinate bias effect.
Journal of Cognitive Psychology, 25, 665–681.
Liddle, W. (1965). An investigation of the Wood Reading
Dynamics method (Doctoral dissertation). University of
Delaware, Newark.
Llewellyn-Thomas, E. (1962). Eye movements in speed reading.
In R. G. Stauffer (Ed.), Speed reading: Practices and procedures (pp. 104–114). Newark, DE: University of Delaware,
Reading Study Center.
Luchette, C. (2014, March 7). Spritz technology promises the
future of reading, so you'll actually read ‘Infinite Jest’ now.
Retrieved from http://www.bustle.com/articles/17529spritz-technology-promises-the-future-of-reading-so-youllactually-read-infinite-jest-now
MacLeod, C. M. (1991). Half a century of research on the Stroop
effect: An integrative review. Psychological Bulletin, 109,
163–203.
32
Masson, M. E. J. (1982). Cognitive processes in skimming
stories. Journal of Experimental Psychology: Learning,
Memory, and Cognition, 8, 400–417.
Masson, M. E. J. (1983). Conceptual processing of text during skimming and rapid sequential reading. Memory &
Cognition, 11, 262–274.
Masson, M. E. J. (1986). Comprehension of rapidly presented
sentences: The mind is quicker than the eye. Journal of
Memory and Language, 25, 588–604.
Matin, E. (1974). Saccadic suppression: A review and an analysis. Psychological Bulletin, 81, 899–917.
McConkie, G. W., & Rayner, K. (1975). The span of the effective stimulus during a fixation in reading. Perception &
Psychophysics, 17, 578–586.
McConkie, G. W., & Rayner, K. (1976). Asymmetry of the
perceptual span in reading. Bulletin of the Psychonomic
Society, 8, 365–368.
McConkie, G. W., & Zola, D. (1979). Is visual information integrated across successive fixations in reading? Perception &
Psychophysics, 25, 221–224.
McLaughlin, G. H. (1969). Reading at “impossible” speeds.
Journal of Reading, 12, 449–510.
Miellet, S., O’Donnell, P. J., & Sereno, S. C. (2009). Parafoveal
magnification: Visual acuity does not modulate the perceptual span in reading. Psychological Science, 20, 721–728.
Mitchell, D. C. (1979). The locus of experimental effects in the
rapid serial visual presentation (RSVP) task. Perception &
Psychophysics, 25, 143–149.
Mitchell, D. C. (1984). An evaluation of subject-paced reading
tasks and other methods for investigating immediate processes in reading. In D. E. Kieras & M. A. Just (Eds.), New
methods in reading comprehension research (pp. 69–89).
Hillsdale, NJ: Erlbaum.
Morrison, R. E., & Rayner, K. (1981). Saccade size in reading depends upon character spaces and not visual angle.
Attention, Perception, & Psychophysics, 30, 395–396.
O’Regan, J. K., Lévy-Schoen, A., Pynte, J., & Brugaillère, B. É. (1984).
Convenient fixation location within isolated words of different length and structure. Journal of Experimental Psychology:
Human Perception and Performance, 10, 250–257.
Perret, P., & Ducrot, S. (2010). Viewing-position effects in
the Stroop task: Initial fixation position modulates Stroop
effects in fully colored words. Psychonomic Bulletin &
Review, 17, 550–555.
Petrick, M. S. (1981). Acoustic and semantic encoding during
rapid reading (Doctoral dissertation). Retrieved from UMI
Dissertation Express. (0356885)
Piantadosi, S. T., Tily, H., & Gibson, E. (2012). The communicative function of ambiguity in language. Cognition, 122,
280–291.
Pinker, S. (1997). Foreword. In D. McGuinness (Ed.), Why our
children can’t read and what we can do about it (pp. ix–x).
New York, NY: Free Press.
Pollatsek, A., Bolozky, S., Well, A. D., & Rayner, K. (1981).
Asymmetries in the perceptual span for Israeli readers.
Brain & Language, 14, 174–180.
Pollatsek, A., Lesch, M., Morris, R. K., & Rayner, K. (1992).
Phonological codes are used in integrating information
across saccades in word identification and reading. Journal
Rayner et al.
of Experimental Psychology: Human Perception and
Performance, 18, 148–162.
Pollatsek, A., Raney, G. E., Lagasse, L., & Rayner, K. (1993). The
use of information below fixation in reading and in visual
search. Canadian Journal of Experimental Psychology, 47,
179–200.
Pollatsek, A., Tan, L. H., & Rayner, K. (2000). The role of
phonological codes in integrating information across saccadic eye movements in Chinese character identification.
Journal of Experimental Psychology: Human Perception
and Performance, 26, 607–633.
Potter, M. C. (1984). Rapid serial visual presentation (RSVP):
A method for studying language processing. In D. Kieras
& M. Just (Eds.), New methods in reading comprehension
research (pp. 91–118). Hillsdale, NJ: Erlbaum.
Potter, M. C., Kroll, J. F., & Harris, C. (1980). Comprehension
and memory in rapid sequential reading. In R. S. Nickerson
(Ed.), Attention and performance (Vol. VIII, pp. 395–418).
Hillsdale, NJ: Erlbaum.
Potter, M. C., Moryadas, A., Abrams, I., & Noel, A. (1993).
Word perception and misperception in context. Journal
of Experimental Psychology: Learning, Memory, and
Cognition, 19, 3–22.
Potter, M. C., Stiefbold, D., & Moryadas, A. (1998). Word selection in reading sentences: Preceding versus following
contexts. Journal of Experimental Psychology: Learning,
Memory, and Cognition, 24, 68–100.
Radach, R., Huestegge, L., & Reilly, R. (2008). The role of global
top-down factors in local eye-movement control in reading.
Psychological Research, 72, 675–688.
Radach, R., Inhoff, A. W., Glover, L., & Vorstius, C. (2013).
Contextual constraint and N + 2 preview effects in reading. Quarterly Journal of Experimental Psychology, 66,
619–633.
Rayner, K. (1975). The perceptual span and peripheral cues in
reading. Cognitive Psychology, 7, 65–81.
Rayner, K. (1998). Eye movements in reading and information
processing: 20 years of research. Psychological Bulletin,
124, 372–422.
Rayner, K. (2009). The thirty-fifth Sir Frederick Bartlett lecture:
Eye movements and attention in reading, scene perception, and visual search. Quarterly Journal of Experimental
Psychology, 62, 1457–1506.
Rayner, K. (2014). The gaze-contingent moving window in
reading: Development and review. Visual Cognition, 22,
242–258.
Rayner, K., Balota, D. A., & Pollatsek, A. (1986). Against parafoveal semantic preprocessing during eye fixations in reading. Canadian Journal of Psychology/Revue Canadienne de
Psychologie Expérimentale, 40, 473–483.
Rayner, K., & Bertera, J. H. (1979). Reading without a fovea.
Science, 206, 468–469.
Rayner, K., Carlson, M., & Frazier, L. (1983). The interaction
of syntax and semantics during sentence processing: Eye
movements in the analysis of semantically biased sentences. Journal of Verbal Learning and Verbal Behavior,
22, 358–374.
Rayner, K., & Duffy, S. A. (1986). Lexical complexity and fixation times in reading: Effects of word frequency, verb
How Do We Read, and Can Speed Reading Help?
complexity and lexical ambiguity. Memory & Cognition, 14,
191–201.
Rayner, K., Juhasz, B. J., & Brown, S. J. (2007). Do readers obtain
preview benefit from word n + 2? A test of serial attention shift
versus distributed lexical processing models of eye movement control in reading. Journal of Experimental Psychology:
Human Perception and Performance, 33, 230–245.
Rayner, K., Liversedge, S. P., & White, S. J. (2006). Eye movements
when reading disappearing text: The importance of the word
to the right of fixation. Vision Research, 46, 310–323.
Rayner, K., Liversedge, S. P., White, S. J., & Vergilino-Perez, D.
(2003). Reading disappearing text: Cognitive control of eye
movements. Psychological Science, 14, 385–388.
Rayner, K., McConkie, G. W., & Zola, D. (1980). Integrating
information across eye movements. Cognitive Psychology,
12, 206–226.
Rayner, K., & Morrison, R. E. (1981). Eye movements and
identifying words in parafoveal vision. Bulletin of the
Psychonomic Society, 17, 135–138.
Rayner, K., & Reingold, E. M. (2015). Evidence for direct cognitive control of fixation durations during reading. Current
Opinion in Behavioral Sciences, 1, 107–112.
Rayner, K., & Schotter, E. R. (2014). Semantic preview benefit
in reading English: The effect of initial letter capitalization.
Journal of Experimental Psychology: Human Perception
and Performance, 40, 1617–1628.
Rayner, K., Schotter, E. R., & Drieghe, D. (2014). Lack of
semantic parafoveal preview benefit in reading revisited.
Psychonomic Bulletin & Review, 21, 1067–1072.
Rayner, K., Slowiaczek, M. L., Clifton, C., & Bertera, J. H.
(1983). Latency of sequential eye movements: Implications
for reading. Journal of Experimental Psychology: Human
Perception and Performance, 9, 912–922.
Rayner, K., Well, A. D., & Pollatsek, A. (1980). Asymmetry of the
effective visual field in reading. Perception & Psychophysics,
27, 537–544.
Reicher, G. M. (1969). Perceptual recognition as a function of meaningfulness of stimulus material. Journal of
Experimental Psychology, 81, 275–280.
Reichle, E. D., Liversedge, S. P., Drieghe, D., Blythe, H. I.,
Joseph, H. S., White, S. J., & Rayner, K. (2013). Using E-Z
Reader to examine the concurrent development of eyemovement control and reading skill. Developmental Review,
33, 110–149.
Reichle, E. D., Pollatsek, A., Fisher, D. L., & Rayner, K. (1998).
Toward a model of eye movement control in reading.
Psychological Review, 105, 125–157.
Reichle, E. D., Rayner, K., & Pollatsek, A. (2003). The
E-Z Reader model of eye-movement control in reading: Comparisons to other models. Behavioral & Brain
Sciences, 26, 445–476.
Reingold, E. M., Reichle, E. D., Glaholt, M. G., & Sheridan, H.
(2012). Direct lexical control of eye movements in reading: Evidence from a survival analysis of fixation durations.
Cognitive Psychology, 65, 177–206.
Sailor, A. L., & Ball, S. E. (1975). Peripheral vision training in
reading speed and comprehension. Perceptual & Motor
Skills, 41, 761–762.
33
Schotter, E. R. (2013). Synonyms provide semantic preview
benefit in English. Journal of Memory and Language, 69,
619–633.
Schotter, E. R., Angele, B., & Rayner, K. (2012). Parafoveal processing in reading. Attention, Perception, & Psychophysics,
74, 5–35.
Schotter, E. R., Bicknell, K., Howard, I., Levy, R., & Rayner, K.
(2014). Task effects reveal cognitive flexibility responding to frequency and predictability: Evidence from eye
movements in reading and proofreading. Cognition, 131,
1–27.
Schotter, E. R., Lee, M., Reiderman, M., & Rayner, K. (2015).
The effect of contextual constraint on parafoveal processing in reading. Journal of Memory and Language, 83,
118–139.
Schotter, E. R., Tran, R., & Rayner, K. (2014). Don’t believe
what you read (only once): Comprehension is supported
by regressions during reading. Psychological Science, 25,
1218–1226.
Schuberth, R. E., & Eimas, P. D. (1977). Effects of context
on the classification of words and nonwords. Journal
of Experimental Psychology: Human Perception and
Performance, 3, 27–36.
Seidenberg, M. S., & McClelland, J. L. (1989). A distributed,
developmental model of word recognition and naming.
Psychological Review, 96, 523–568.
Slattery, T. J. (2009). Word misperception, the neighbor frequency effect, and the role of sentence context: Evidence
from eye movements. Journal of Experimental Psychology:
Human Perception and Performance, 35, 1969–1975.
Slattery, T. J., Angele, B., & Rayner, K. (2011). Eye movements
and display change detection during reading. Journal
of Experimental Psychology: Human Perception and
Performance, 37, 1924–1938.
Slowiaczek, M. L., & Clifton, C., Jr. (1980). Subvocalization
and reading for meaning. Journal of Verbal Learning and
Verbal Behavior, 19, 573–582.
Sparrow, L., & Miellet, S. (2002). Activation of phonological
codes during reading: Evidence from errors detection and
eye movements. Brain & Language, 81, 509–516.
Stanners, R. F., Jastrzembski, J. E., & Westbrook, A. (1975).
Frequency and visual quality in a word-nonword classification task. Journal of Verbal Learning and Verbal Behavior,
14, 259–264.
Stanovich, K. E., & West, R. F. (1979). Mechanisms of sentence
context effects in reading: Automatic activation and conscious attention. Memory & Cognition, 7, 77–85.
Stanovich, K. E., & West, R. F. (1981). The effect of sentence
context on ongoing word recognition: Tests of a two-process theory. Journal of Experimental Psychology: Human
Perception and Performance, 7, 658–672.
Stanovich, K. E., & West, R. F. (1983). On priming by a sentence
context. Journal of Experimental Psychology: General, 112,
1–36.
Stroop, J. R. (1935). Studies of interference in serial verbal reactions. Journal of Experimental Psychology, 18, 643–662.
Sutz, R. (2009). Speed reading for dummies. Hoboken, NJ: John
Wiley & Sons.
34
Taylor, S. E. (1962). An evaluation of forty-one trainees who
had recently completed the “Reading Dynamics” program.
In E. P. Bliesmer & R. C. Staiger (Eds.), Problems, programs,
and projects in college adult reading. Eleventh yearbook of
the National Reading Conference (pp. 51–120). Milwaukee,
WI: National Reading Conference.
Taylor, S. E. (1965). Eye movements in reading: Facts and fallacies. American Educational Research Journal, 2, 187–202.
The Science. (2015). [Internet archive]. Retrieved from https://
web.archive.org/web/20150913052621/http://www.
spritzinc.com/the-science
Treiman, R., Freyd, J. J., & Baron, J. (1983). Phonological
recoding and use of spelling-sound rules in reading of sentences. Journal of Verbal Learning and Verbal Behavior,
22, 682–700.
Underwood, G., Hubbard, A., & Wilkinson, H. (1990). Eye
fixations predict reading comprehension: The relationships
between reading skill, reading speed, and visual inspection. Language and Speech, 33, 69–81.
Van Orden, G. C. (1987). A ROWS is a ROSE: Spelling, sound,
and reading. Memory & Cognition, 15, 181–198.
Vitu, F., O’Regan, J. K., & Mittau, M. (1990). Optimal landing
position in reading isolated words and continuous text.
Perception & Psychophysics, 47, 583–600.
von der Malsburg, T., & Vasishth, S. (2011). What is the scanpath signature of syntactic reanalysis? Journal of Memory
and Language, 65, 109–127.
von der Malsburg, T., & Vasishth, S. (2013). Scanpaths reveal
syntactic underspecification and reanalysis strategies.
Language and Cognitive Processes, 28, 1545–1578.
Walton, H. N. (1957). Vision and rapid reading. American
Journal of Optometry and Archives of American Academy
of Optometry, 34, 73–82.
'War and Peace' in 20 Minutes? If You Care What It Says,
Read [Letter to the editor]. (1995, September 3). The
New York Times. Retrieved from http://www.nytimes.
com/1995/09/03/opinion/l-war-and-peace-in-20-minutesif-you-care-what-it-says-read-449395.html
Rayner et al.
Warren, T., White, S. J., & Reichle, E. D. (2009). Investigating
the causes of wrap-up effects: Evidence from eye movements and E–Z Reader. Cognition, 111, 132–137.
West, R. F., & Stanovich, K. E. (1982). Source of inhibition in
experiments on the effect of sentence context on word
recognition. Journal of Experimental Psychology: Learning,
Memory, and Cognition, 8, 385–399.
Wheeler, D. D. (1970). Processes in word recognition. Cognitive
Psychology, 1, 59–85.
Why Spritz Works: It’s All About the Alignment of Words.
(2014, February 16). Why Spritz works: It’s all about the
alignment of words [Blog post, Internet archive]. Retrieved
from https://web.archive.org/web/20150321193530/http://
www.spritzinc.com/why-spritz-works-its-all-about-thealignment-of-words/
Wilkinson, S. C., Reader, W., & Payne, S. J. (2012). Adaptive
browsing: Sensitivity to time pressure and task difficulty.
International Journal of Human-Computer Studies, 70,
14–25.
World Speed Reading Council. (2008). [Website]. Retrieved
from
http://www.mentalworldrecords.com/worldspeedreadingcouncil/
World’s Fastest Reader on Pelosi Bill [Video file]. (2011, May 5).
Retrieved from http://video.foxbusiness.com/v/3898543/
worlds-fastest-reader-on-pelosi-bill/?#sp=show-clips
Wotschack, C., & Kliegl, R. (2013). Reading strategy modulates parafoveal-on-foveal effects in sentence reading.
Quarterly Journal of Experimental Psychology, 66, 549–
562.
Xu, Y., Pollatsek, A., & Potter, M. C. (1999). The activation
of phonology during silent Chinese word reading. Journal
of Experimental Psychology: Learning, Memory, and
Cognition, 25, 838–857.
Zola, D. (1984). Redundancy and word perception during reading. Perception & Psychophysics, 36, 277–284.
Zwaan, R. A., & Radvansky, G. A. (1998). Situation models
in language comprehension and memory. Psychological
Bulletin, 123, 162–185.